From 487d028ac48ee415eca298b409c74f1a299edf3c Mon Sep 17 00:00:00 2001 From: Mahmoud Ahmed Ali <mahmoudali2929@gmail.com> Date: Wed, 1 Mar 2023 11:49:48 +0000 Subject: [PATCH] Add demo file --- demo_new.ipynb | 2051 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 2051 insertions(+) create mode 100644 demo_new.ipynb diff --git a/demo_new.ipynb b/demo_new.ipynb new file mode 100644 index 0000000..9c21a2b --- /dev/null +++ b/demo_new.ipynb @@ -0,0 +1,2051 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# PVNet 6D Pose PipeLine Demo\n", + "\n", + "## Introduction\n", + "\n", + "In this notebook we'll be performing a quick version of the 6D pose estimation solution implemented by this project. The process is similar to the flow contained in an iteration of the pipeLine.py evalModels function." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Contributions\n", + "\n", + "* Propose a novel framework for 6D pose estimation using a pixel-wise voting network (PVNet), which learns a vector-field representation for robust 2D keypoint localization and naturally deals with occlusion and truncation.\n", + "\n", + "* Propose to utilize an uncertainty-driven PnP algorithm to account for uncertainties in 2D keypoint localizations, based on the dense predictions from PVNet.\n", + "\n", + "* Demonstrate significant performance improvements of our approach compared to the state of the art on benchmark datasets (ADD: 86.3% vs. 79% and 40.8% vs. 30.4% on LINEMOD and OCCLUSION, respectively). \n", + "\n", + "* Create a new dataset (Truncation LINEMOD) for evaluation on truncated objects.\n", + "\n", + "* Real time pose estimation with 25 fps." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## PVNet Architecture\n", + "\n", + "" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Proposed approch [PVNet]\n", + "\n", + "- Two stage pipeline Methods:\n", + "1. First detect 2D object keypoints using CNN's features.\n", + "2. Compute 6D pose parameters using the PnP algorithm.\n", + "\n", + "- **Innovation for PVNet**: is in a new representation for 2D object keypoints as well as a modified PnP algorithm for pose estimation.\n", + "\n", + " - PVNet method uses a Pixel-wise Voting Network (PVNet) to detect 2D keypoints in a RANSAC-like fashion, which robustly handles occluded and truncated objects. \n", + " - The RANSAC-based voting also gives a spatial probability distribution of each keypoint, allowing us to estimate the 6D pose with an uncertainty-driven PnP." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Data Format\n", + "* Images: RGB images (H, W, 3)\n", + "* Mask: mask imgage (H, w, 1)\n", + "* Labels: .txt file \n", + " * 9 bbox 2d pixel location of a set of 3d keypoints.\n", + " * Format: {Object tag}{x1}{y1}......{x9}{y9}.\n", + " * Values: [0-1] given relative coordinate on image ex: x1=0.1, y1=0.5, img=[640,480] => pixel(64,240)\n", + "* 3D keypoint: .txt file\n", + " * File containing corresponding keypoint locations in 3d coordinate system (for Pnp)." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Read image and labels" + ] + }, + { + "cell_type": "code", + "execution_count": 52, + "metadata": { + "scrolled": false + }, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "RGB Image Shape: (480, 640, 3)\n", + "Keypoint Labels: ['6.686142628874398586e-01', '7.263400377314979117e-01', '5.493648993389658175e-01', '9.312443868540029035e-01', '5.769284322293616318e-01', '7.881035621388378098e-01', '6.407135761041395927e-01', '8.899895375398150232e-01', '5.877399934822387095e-01', '7.999863116729838408e-01', '5.829383316205479781e-01', '9.211855263900207147e-01', '6.211462113671539775e-01', '7.124319717426675913e-01', '6.357183510471959842e-01', '8.133651672576241998e-01', '6.086036276270714307e-01', '8.212373153805835324e-01\\n']\n", + "# Keypoint Labels: 18\n" + ] + }, + { + "data": { + "image/png": "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\n", + "text/plain": [ + "<Figure size 432x288 with 1 Axes>" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "import sys\n", + "\n", + "import tensorflow as tf, numpy as np, matplotlib.pyplot as plt, random, math, statistics, os\n", + "from skimage.io import imshow\n", + "import data\n", + "import numpy as np\n", + "\n", + "class_name = 'pear'\n", + "\n", + "image, labels, mask, ch, val = data.getDataSplitImage(True, modelClass=class_name) # Loading a random image from the validation data set\n", + "\n", + "print(\"RGB Image Shape: \" + str(image.shape)) # an rgb image of dimensions 480 x 640 x 3\n", + "print(\"Keypoint Labels: \"+ str(labels)) # a labels file detailing x, y coordinates for 9 2d keypoints on our loaded image\n", + "print(\"# Keypoint Labels: \"+ str(len(labels))) # a labels file detailing x, y coordinates for 9 2d keypoints on our loaded image\n", + "\n", + "def showImage(img): # displays image using plt\n", + "\tplt.figure()\n", + "\timshow(np.squeeze(img))\n", + "\tplt.show()\n", + " \n", + "showImage(image)" + ] + }, + { + "cell_type": "code", + "execution_count": 53, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "['4294.png', '4294.png', '4294.txt']" + ] + }, + "execution_count": 53, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "ch" + ] + }, + { + "cell_type": "code", + "execution_count": 54, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "4294" + ] + }, + "execution_count": 54, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "img_id = ch[0].split('.png')\n", + "int(img_id[0])" + ] + }, + { + "cell_type": "code", + "execution_count": 55, + "metadata": {}, + "outputs": [ + { + "data": { + "image/png": "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\n", + "text/plain": [ + "<Figure size 432x288 with 1 Axes>" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "showImage(mask)" + ] + }, + { + "cell_type": "code", + "execution_count": 56, + "metadata": {}, + "outputs": [], + "source": [ + "# mask = mask /255.0" + ] + }, + { + "cell_type": "code", + "execution_count": 57, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "255" + ] + }, + "execution_count": 57, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "mask.max()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\n", + "We now have our image, and a list containing info for the corresponding 2d bounding box keypoints. Let's load the models we'll be using to generate our predictions for unit vector and class outputs." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Predicte unit vector and class output\n", + "- Assuming **C** classes of objects and **K** keypoints for each class, PVNet takes as input the H × W × 3 image\n", + " * Class output: H × W × (C+1)\n", + " * Unit vector: H × W × (K*2*C)" + ] + }, + { + "cell_type": "code", + "execution_count": 58, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.iter\n", + "WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_1\n", + "WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_2\n", + "WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.decay\n", + "WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.learning_rate\n", + "WARNING:tensorflow:A checkpoint was restored (e.g. tf.train.Checkpoint.restore or tf.keras.Model.load_weights) but not all checkpointed values were used. See above for specific issues. Use expect_partial() on the load status object, e.g. tf.train.Checkpoint.restore(...).expect_partial(), to silence these warnings, or use assert_consumed() to make the check explicit. See https://www.tensorflow.org/guide/checkpoint#loading_mechanics for details.\n", + "WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.iter\n", + "WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_1\n", + "WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.beta_2\n", + "WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.decay\n", + "WARNING:tensorflow:Unresolved object in checkpoint: (root).optimizer.learning_rate\n", + "WARNING:tensorflow:A checkpoint was restored (e.g. tf.train.Checkpoint.restore or tf.keras.Model.load_weights) but not all checkpointed values were used. See above for specific issues. Use expect_partial() on the load status object, e.g. tf.train.Checkpoint.restore(...).expect_partial(), to silence these warnings, or use assert_consumed() to make the check explicit. See https://www.tensorflow.org/guide/checkpoint#loading_mechanics for details.\n", + "Vector Prediction shape: (480, 640, 18)\n", + "Class Prediction shape: (480, 640, 1)\n" + ] + }, + { + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXAAAAEYCAYAAABFvq0IAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8li6FKAAARiUlEQVR4nO3df6jd9X3H8ecrNzFVY9FoE4IRayGMxTKtE5fhKNK1M/0ZixSyspE/HPnHQWWDEids9L+u0lIoOJC2LtCuEqpbRco2yVr8Z2hjq9P4o6bV1WDMna0/h43JzXt/3G+603iT3OSec77nc+/zAYfzPZ/zPef7fus9r3zu557vOakqJEntWdZ3AZKkM2OAS1KjDHBJapQBLkmNMsAlqVEGuCQ1amQBnmRzkmeS7EuyY1THkaSlKqN4H3iSKeCnwEeA/cCPgD+tqieHfjBJWqJGNQO/BthXVT+vqreBu4EtIzqWJC1Jy0f0vBcDLwzc3g/8weAOSbYD27ubvz+iOiSpdS9X1XvmumNUAZ45xn5rraaq7gTuBEji+fySNLf/PtEdo1pC2Q9cMnB7PfDiiI4lSUvSqAL8R8CGJJclOQvYCtw3omNJ0pI0kiWUqjqS5C+BfwOmgG9W1d5RHEuSlqqRvI3wtItwDVySTuSRqrp6rjs8E1OSGmWAS1KjDHBJapQBLkmNMsAlqVEGuCQ1ygCXpEYZ4JLUKANckhplgEtSowxwSWqUAS5JjTLAJalRBrgkNcoAl6RGGeCS1CgDXJIaZYBLUqMMcElqlAEuSY0ywCWpUQa4JDXKAJekRhngktQoA1ySGmWAS1KjDHBJapQBLkmNMsAlqVEGuCQ1ygCXpEYZ4JLUKANckhplgEtSowxwSWqUAS5JjTLAJalRBrgkNcoAl6RGGeCS1CgDXJIaZYBLUqNOGeBJvplkOskTA2OrkzyQ5Nnu+oKB+25Nsi/JM0muH1XhkrTUzWcG/o/A5uPGdgC7q2oDsLu7TZKNwFbg8u4xdySZGlq1kqTfOGWAV9WDwK+OG94C7Oy2dwI3DIzfXVWHquo5YB9wzZBqlSQNONM18LVVdQCgu17TjV8MvDCw3/5uTJI0ZMuH/HyZY6zm3DHZDmwf8vElack40xn4wSTrALrr6W58P3DJwH7rgRfneoKqurOqrq6qq8+wBkla0s40wO8DtnXb24DvDYxvTbIyyWXABuDhhZUoSZrLKZdQknwHuA64KMl+4O+ALwK7ktwE/AL4DEBV7U2yC3gSOALcXFUzI6pdkpa0VM25RD3eIpL+i5CkyfTIiZaaPRNTkhplgEtSowxwSWqUAS5JjTLAJalRBrgkNcoAl6RGGeCS1CgDXJIaZYBLUqMMcElqlAEuSY0ywCWpUQa4JDXKAJekRhngktQoA1ySGmWAS1KjDHBJapQBLkmNMsAlqVEGuCQ1ygCXpEYZ4JLUKANckhplgEtSowxwSWqUAS5JjTLAJalRBrgkNcoAl6RGGeCS1CgDXJIaZYBLUqMMcElqlAEuSY0ywCWpUQa4JDXKAJekRhngktQoA1ySGmWAS1KjDHBJatQpAzzJJUl+kOSpJHuTfK4bX53kgSTPdtcXDDzm1iT7kjyT5PpRNiBJS9V8ZuBHgL+uqt8FNgE3J9kI7AB2V9UGYHd3m+6+rcDlwGbgjiRToyhekpayUwZ4VR2oqh93228ATwEXA1uAnd1uO4Ebuu0twN1VdaiqngP2AdcMu3BJWupOaw08yXuBDwAPAWur6gDMhjywptvtYuCFgYft78aOf67tSfYk2XP6ZUuSls93xySrgHuAW6rq9SQn3HWOsXrHQNWdwJ3dc7/jfknSyc1rBp5kBbPh/e2qurcbPphkXXf/OmC6G98PXDLw8PXAi8MpV5J0zHzehRLgG8BTVfWVgbvuA7Z129uA7w2Mb02yMsllwAbg4eGVLEmC+S2hXAv8OfB4kke7sb8BvgjsSnIT8AvgMwBVtTfJLuBJZt/BcnNVzQy9ckla4lLV//Kza+CSdEKPVNXVc93hmZiS1CgDXJIaZYBLUqMMcElqlAEuSY0ywCWpUQa4JDXKAJekRhngktQoA1ySGmWAa8lYtswfdy0u/kRrSVi9ejVr1qzhiiuu6LsUaWjm/YUOUstee+01ZmZmeO211/ouRRoaZ+BaEmZmZj/R+K233uq5Eml4DHBJapQBrmZNTU2xYsUKpqamAFixYgWrVq1i+XJXBrU0GOBqUhLOPvvs3xo7fPgwhw8f/s1yibTYOVVRk6qKN9988x3jhw4d6qEaqR/OwCWpUQa4JsrU1BSbNm3ypBtpHvxSY02UJCTh6NGjfZciTYoTfqmxa+CaKFXFJEwqpBb4e6okNcoAl6RGGeCS1CgDXE1K0ncJUu8McDXJP3RKBrgkNcsAl6RGGeCS1CgDXJIaZYBLUqMMcElqlAEuSY0ywCWpUQa4JDXKAJekRhngktQoA1ySGmWAS1KjDHBJapQBLkmNOmWAJ3lXkoeTPJZkb5IvdOOrkzyQ5Nnu+oKBx9yaZF+SZ5JcP8oGJGmpms8M/BDwoaq6ArgS2JxkE7AD2F1VG4Dd3W2SbAS2ApcDm4E7kkyNonhJWspOGeA1683u5oruUsAWYGc3vhO4odveAtxdVYeq6jlgH3DNUKuWJM1vDTzJVJJHgWnggap6CFhbVQcAuus13e4XAy8MPHx/N3b8c25PsifJnoU0IE0qv7dTozavAK+qmaq6ElgPXJPk/SfZfa6f2nd8gWFV3VlVV1fV1fMrVWrL1JQrhxqt03oXSlW9CvyQ2bXtg0nWAXTX091u+4FLBh62HnhxwZVKjbnuuuv6LkGL3HzehfKeJOd322cDHwaeBu4DtnW7bQO+123fB2xNsjLJZcAG4OFhFy5NuldffdVlFI3U8nnssw7Y2b2TZBmwq6ruT/KfwK4kNwG/AD4DUFV7k+wCngSOADdX1cxoypcm1+OPP07VO1YPpaHJJPyAJem/CEmaTI+c6G+FnokpSY0ywCWpUQa4JDXKAJekRhngktQoA1ySGmWAS1KjDHBpEauq31xmZjyfbrExwKVFZjC0By1btoyq4o033uipMg2bAS4tAocOHZoztOeyatUqbrvttjFUpVHzVHqpYUePHj3jD8zyg7aa4an00mJybLa9kBD+9a9/PcSK1If5fBqhpAkxzN+YV65cObTnUj+cgUsNuPTSS4f+0bSTsHyqhXEGLk24UQXt+eefP5Ln1fg4A5cm2Chnya+//vrInlvjYYBLE2qU4e0XLi8OBrg0YR588MGRr08fPXp0pM+v8XANXJogR44cGfnseO3atSN9fo2PAS5NiIWclHM6pqenR34MjYcBLk2Acb2lz7MvFxfXwKWejSu8b7/99rEcR+PjZ6FIPRrn68/Zd7P8LBRp0tx1111jO5bhvTg5A5d6Mq7X3uHDhznrrLPGciyNhDNwaZKMc+K0Zs2asR1L42WAS2O2bdu2sR7v1VdfHevxND4uoUhjdO655/Lmm2+O7XjnnHMOb7311tiOp5FwCUWaBOMMb8DwXuQMcGlMxv3b7ve///2xHk/j5xKKNGJ9vcZ86+Ci4RKK1Ievfe1rvRzXTxtcGpyBSyPk7FtD4AxcGrdJmBxpcTPApRHocwlj2TJf1kuF/6elEehzCcOZ/9JhgEtD1meA+r7vpcUAlxaRc845p+8SNEYGuDRELl9onAxwSWqUAS4NyZe//OW+S9AS44k80pBMyGup7xI0fAs/kSfJVJKfJLm/u706yQNJnu2uLxjY99Yk+5I8k+T6hdcvSTre6SyhfA54auD2DmB3VW0Adne3SbIR2ApcDmwG7kgyNZxypcn08ssv912ClqB5BXiS9cDHga8PDG8BdnbbO4EbBsbvrqpDVfUcsA+4ZjjlSpPpwgsv7LuEiVjC0XjNdwb+VeDzwOD5wWur6gBAd33si/cuBl4Y2G9/N/ZbkmxPsifJntOuWtI7HD58uO8SNGanDPAknwCmq+qReT7nXH9FecfUoKrurKqrT7Q4L+n0vPTSS32XoDFbPo99rgU+leRjwLuAdyf5FnAwybqqOpBkHTDd7b8fuGTg8euBF4dZtDRJXnnllb5LAODpp5/uuwSNW1XN+wJcB9zfbd8O7Oi2dwBf6rYvBx4DVgKXAT8Hpk7xvOXFS6uXSbFx48be/1t4GcllT50gOxdyIs8XgY8keRb4SHebqtoL7AKeBP4VuLmqZhZwHEnzcNVVV/VdgsbME3mkBZqE1xDMfuP9eeed13cZGj6/kUeSFhsDXFokDh061HcJGjMDXFoknn/++b5L0JgZ4NIisW7dur5L0JgZ4NICTE1Nzsf83HjjjX2XoDHzXSjSAkxNTXHkyJG+ywD8KNlFzHehSKMwM+MpDuqPAS5JjTLAJalRBrgkNcoAl6RGGeDSIrBp06a+S1APDHBpgX75y1/2XQIPPfRQ3yWoBwa4tEAXXXRR3yVoiTLApca9/fbbfZegnhjgUuNuueWWvktQTzyVXhqCNWvWcPDgwV6O7Sn0i56n0kujND09feqdpCEzwKUhueeee8Z+TGffS5tLKNIQjfv1ZIAvCS6hSOMwzkA1vGWAS0M2jmD99Kc/PfJjaPIt77sAaTFKwszMDMuWDX+O5MxbxzgDl0ZkamqKe++9d6jP+corrwz1+dQ2A1waoRtvvJFPfvKTQ/vj5urVq4fyPFocfBeKNCYLfa25dLJk+S4UqW9JSHLaX4L82c9+1vDWnPwjpjRmK1asAOY3Ize4dTLOwKWeHJuRH+/1118/4X3SIGfgUs8Map0pZ+CS1CgDXJIaZYBLUqMMcElqlAEuSY2alHehvAz8b3e9GFyEvUwie5lMi6kXGH4/l57ojok4lR4gyZ4TnS7aGnuZTPYymRZTLzDeflxCkaRGGeCS1KhJCvA7+y5giOxlMtnLZFpMvcAY+5mYNXBJ0umZpBm4JOk0GOCS1KjeAzzJ5iTPJNmXZEff9ZxKkm8mmU7yxMDY6iQPJHm2u75g4L5bu96eSXJ9P1XPLcklSX6Q5Kkke5N8rhtvrp8k70rycJLHul6+0I0318sxSaaS/CTJ/d3tlnt5PsnjSR5Nsqcba7KfJOcn+W6Sp7vXzh/21ktV9XYBpoCfAe8DzgIeAzb2WdM8av4gcBXwxMDYl4Ad3fYO4O+77Y1dTyuBy7pep/ruYaDudcBV3fZ5wE+7mpvrBwiwqtteATwEbGqxl4Ge/gr4J+D+ln/OuhqfBy46bqzJfoCdwF9022cB5/fVS98z8GuAfVX186p6G7gb2NJzTSdVVQ8CvzpueAuz/1Pprm8YGL+7qg5V1XPAPmZ7nghVdaCqftxtvwE8BVxMg/3UrDe7myu6S9FgLwBJ1gMfB74+MNxkLyfRXD9J3s3sJO4bAFX1dlW9Sk+99B3gFwMvDNze3421Zm1VHYDZUATWdOPN9JfkvcAHmJ25NtlPt+TwKDANPFBVzfYCfBX4PHB0YKzVXmD2H9N/T/JIku3dWIv9vA/4H+Cubnnr60nOpade+g7wub6KZDG9r7GJ/pKsAu4Bbqmq10+26xxjE9NPVc1U1ZXAeuCaJO8/ye4T20uSTwDTVfXIfB8yx9hE9DLg2qq6CvgocHOSD55k30nuZzmzS6j/UFUfYPYznE72t7uR9tJ3gO8HLhm4vR54sadaFuJgknUA3fV0Nz7x/SVZwWx4f7uq7u2Gm+0HoPuV9ofAZtrs5VrgU0meZ3ZZ8UNJvkWbvQBQVS9219PAPzO7jNBiP/uB/d1vdwDfZTbQe+ml7wD/EbAhyWVJzgK2Avf1XNOZuA/Y1m1vA743ML41ycoklwEbgId7qG9Omf0yxm8AT1XVVwbuaq6fJO9Jcn63fTbwYeBpGuylqm6tqvVV9V5mXxP/UVV/RoO9ACQ5N8l5x7aBPwGeoMF+quol4IUkv9MN/THwJH31MgF/0f0Ys+9++BlwW9/1zKPe7wAHgMPM/ut6E3AhsBt4trtePbD/bV1vzwAf7bv+43r5I2Z/nfsv4NHu8rEW+wF+D/hJ18sTwN924831clxf1/H/70Jpshdm140f6y57j73OG+7nSmBP97P2L8AFffXiqfSS1Ki+l1AkSWfIAJekRhngktQoA1ySGmWAS1KjDHBJapQBLkmN+j//GP+D2HfxXwAAAABJRU5ErkJggg==\n", + "text/plain": [ + "<Figure size 432x288 with 1 Axes>" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "import models\n", + "\n", + "\n", + "nnInput = np.array([image])\n", + "\n", + "# loading our model to predict unit vectors per pixel per keypoint on image\n", + "vecModel = models.stvNetNew(outVectors = True, outClasses = False)\n", + "vecModel.load_weights(f'models/stvNet_new_coords_{class_name}') # loading weights for standard labels model\n", + "vecModel.compile(optimizer = tf.keras.optimizers.Adam(), loss = tf.keras.losses.Huber())\n", + "\n", + "# loading our class model for image segmentation\n", + "classModel = models.uNet(outVectors = False, outClasses = True)\n", + "classModel.load_weights(f'models/uNet_classes_{class_name}')\n", + "classModel.compile(optimizer = tf.keras.optimizers.Adam(), loss = tf.keras.losses.BinaryCrossentropy())\n", + "\n", + "vecPred = vecModel.predict(nnInput)[0]\n", + "classPred = classModel.predict(nnInput)[0]\n", + "\n", + "print(\"Vector Prediction shape: \" + str(vecPred.shape))\n", + "print(\"Class Prediction shape: \" + str(classPred.shape))\n", + "showImage(classPred) # let's see our class prediction output" + ] + }, + { + "cell_type": "code", + "execution_count": 59, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "1.0" + ] + }, + "execution_count": 59, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "classPred.max()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Model Prediction Results\n", + "\n", + "We can see that our vector prediction has a shape that matches our photo dimensions in width and height, but has 18 values for each pixel. Each value represents an x or y component for our unit vector predictions for each of the 9 keypoints we are looking for.\n", + "\n", + "Our class prediction has (hopefully) found the pixels associated with our object of interest. We can now filter for these pixels and use the associated vector predictions for our ransac voting process." + ] + }, + { + "cell_type": "code", + "execution_count": 60, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "7588\n" + ] + } + ], + "source": [ + "population = np.where(classPred > .9)[:2] #.9\n", + "population = list(zip(population[0], population[1]))\n", + "\n", + "print(len(population)) # the number of class pixels found" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## RanSaC Voting\n", + "\n", + "The goal of RanSaC voting algorithm:\n", + "- **Input:** the predictions on the pixels of interest.\n", + "- **Output:** generate a list of hypotheses, and an associated weight for each hypothesis (which indicates how well the prediction 'fits' the rest of the data).\n", + "\n", + "1. Take two pixels from our population at random. \n", + "2. For each keypoint, find the intersection of the lines given by the unit vector predictions for each keypoint, which will be used as our hypothesis. \n", + "\n", + " - This process also involves a fairly intimidating-looking if statement, which checks that our hypothesis was generated on the correct side of the two queried pixels according to their unit vector predictions.\n", + "\n", + "- We then see how well this hypothesis aligns with the rest of the population. \n", + "\n", + "3. For every other pixel of interest, we see find the unit vector between our hypothesis and the pixel, and see how similar the real unit vector is to the unit vector predicted by our vector model. \n", + "\n", + " - Taking the dot product of the two unit vectors, which will have a value between 1 and -1. \n", + " - 1 for unit vectors pointing in the same direction. \n", + " - -1 for the opposite direction.\n", + "\n", + " - If this value is above a certain threshold, in this case .99, increment the weight value for the hypothesis. \n", + " - Use this weight value to determine which hypotheses best fit the rest of the population data.\n", + " - Higher voting score means that a hypothesis is more confident as it coincides with more predicted directions.\n", + " \n", + " - The resulting hypotheses characterize the spatial probability distribution of a keypoint in the image. \n", + " - Finally, the mean and the covariance for a keypoint are estimated." + ] + }, + { + "cell_type": "code", + "execution_count": 61, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "(364, 392) (417, 390)\n", + "[ 0.9219574 -0.41446587 -0.46761122 0.89711267 -0.8471135 0.49623317\n", + " 0.25534162 0.938099 -0.6129987 0.75820845 -0.26452094 0.9616415\n", + " 0.25848803 -1.0394933 0.48156703 0.82792467 -0.06503494 0.9682 ] [ 4.5332789e-01 -9.2610180e-01 -8.1497896e-01 5.9311306e-01\n", + " -4.6418869e-01 -9.3715292e-01 8.6694086e-01 4.7118798e-01\n", + " -3.8097799e-01 -9.3335295e-01 -5.8838826e-01 8.0427396e-01\n", + " 8.5847110e-02 -1.0268021e+00 5.4507983e-01 -8.7991989e-01\n", + " -8.8869863e-05 -9.9536514e-01]\n", + "350.1992483074778 422.69904000881047\n", + "447.04791757485043 348.71208325073934\n", + "376.82772465231636 370.10195026112837\n", + "427.47779527463376 409.2780517344803\n", + "383.4249279607432 376.2952740522515\n", + "444.55466589268417 369.84164175101273\n", + "349.27130754264795 395.6625444689548\n", + "389.6690731675227 406.930560189549\n", + "393.805756963814 389.9979182405537\n", + "--------------------\n", + "(335, 410) (449, 399)\n", + "[ 0.83968586 0.52475053 -0.4507978 0.8927477 -0.6863344 0.7134837\n", + " -0.03030851 0.98045754 -0.55777407 0.81809074 -0.34335497 0.95857483\n", + " -0.8769559 0.4015841 -0.03982239 0.95932937 -0.31317708 0.89241797] [ 0.2747072 -1.0247124 -1.0141933 -0.0702555 -0.35760936 -0.96135837\n", + " 0.48431545 -0.8845623 -0.32861578 -0.9800562 -1.0057633 -0.30342448\n", + " 0.01707656 -1.0290505 0.11278641 -1.0442634 -0.14363207 -0.9782076 ]\n", + "345.4704516491785 426.75441952985426\n", + "445.8833951391574 354.008827078512\n", + "375.03661511441663 371.48684884084736\n", + "434.528630258832 406.92330957232053\n", + "383.39679155973204 377.00307869841436\n", + "440.87726740150674 372.07548377160776\n", + "339.20286014743795 400.8220389234738\n", + "354.74059252838924 409.1805550081048\n", + "390.7270136437048 390.44366734946993\n", + "--------------------\n", + "(379, 364) (380, 354)\n", + "[ 0.96650326 -0.44970274 -0.22808957 0.967167 0.9532856 -0.22364953\n", + " 0.6346556 0.7230317 0.979081 0.36271966 0.15021698 0.98851275\n", + " 0.66499513 -0.7711252 0.91206676 0.17282197 0.9203675 0.47995046] [ 0.92686415 -0.38662803 -0.07835028 1.01688 0.99329346 -0.08255363\n", + " 0.68717164 0.6258658 1.0089035 0.15782425 0.2524723 0.9565756\n", + " 0.8614431 -0.7590806 0.9646804 0.03308013 1.0507816 0.33092827]\n", + "348.35597040062885 429.86029550564785\n", + "441.4938759638751 349.26190348864293\n", + "429.4031830526439 408.24240525268516\n", + "383.43846514179666 375.9806489207281\n", + "470.6661168040091 377.9298227215221\n", + "346.4650516873919 392.0571589804388\n", + "380.63963894850224 372.65318364250874\n", + "389.47630158994514 384.0896720486172\n", + "--------------------\n", + "(367, 380) (437, 356)\n", + "[ 0.94019073 -0.37883916 -0.36168346 0.9292867 -0.6806555 0.70071477\n", + " 0.4150925 0.86004627 -0.21775642 0.94179684 -0.13501874 1.0029095\n", + " 0.5701352 -0.8826507 0.73987293 0.62442654 0.3399615 0.9029555 ] [ 0.58512276 -0.77427655 -0.4702765 0.85534745 0.24275596 -0.9984277\n", + " 0.9744894 -0.18676619 0.3237887 -0.94008815 0.90486825 0.32597902\n", + " 0.4390359 -0.9561372 0.7859018 -0.6935123 0.6313599 -0.81317246]\n", + "350.25719540946295 421.5517480113508\n", + "457.2013600646186 344.89313954727663\n", + "376.5852779982043 370.6891189089681\n", + "426.86293425284543 408.8922315693304\n", + "367.9686693894582 379.7760302631709\n", + "442.0082275488344 369.90186455392904\n", + "323.40163366949685 408.161721638178\n", + "390.8666425565587 408.2791994058338\n", + "393.3238110471316 389.9108789780363\n", + "--------------------\n", + "(345, 396) (430, 404)\n", + "[ 1.0378468 0.05993053 -0.42548186 0.938474 -0.61686015 0.76136136\n", + " 0.11347949 0.96008956 -0.40649897 0.8360032 -0.24027696 0.96745473\n", + " 0.48134473 -0.9110181 0.24778557 0.92584914 -0.11650407 0.93076515] [ 0.28877178 -0.933109 -0.97383535 0.25264573 -0.5198798 -0.89574224\n", + " 0.8601572 -0.44501463 -0.48988634 -0.8677709 -0.97564834 0.32303613\n", + " -0.04802652 -1.0556477 0.09137382 -1.0080606 -0.38061002 -0.9891394 ]\n", + "346.9461762922956 429.702904302672\n", + "443.6826161601681 351.2596437686027\n", + "374.7233170258653 371.9179669079716\n", + "429.002050233255 405.9287707644196\n", + "383.0532714296618 377.4969421360154\n", + "440.5020546757712 372.28111912696835\n", + "337.79794156617925 399.8052733949777\n", + "388.8342961435146 407.7313994829838\n", + "393.44910566946555 389.9356153175189\n", + "--------------------\n", + "(381, 395) (466, 373)\n", + "[ 0.7035231 -0.7484307 -0.5800984 0.8264629 -1.0100517 -0.14953618\n", + " 0.267592 0.9593303 -0.9823527 0.13140236 -0.3583996 0.94423515\n", + " 0.08532479 -1.0600996 0.7867156 0.58395964 -0.4000011 0.89380896] [ 0.41156784 -0.92019856 -0.7443626 -0.63815373 -0.03608726 -1.0979465\n", + " 0.73623985 -0.79789454 0.08385985 -0.94529516 -0.02865582 -0.9805832\n", + " 0.22377546 -1.0151474 0.36683312 -0.98091155 0.20532307 -0.9718996 ]\n", + "348.4936986216981 425.5558447624296\n", + "445.84199071212953 349.48708870364806\n", + "377.31138161065996 370.0849905994001\n", + "427.96152134030353 408.0992710876868\n", + "382.9573574030198 380.36696466201823\n", + "440.8938442314459 372.26632968802\n", + "357.68502357311445 396.87656470849106\n", + "386.68656781139083 402.6609950146885\n", + "398.11190675403054 387.3420079995756\n", + "--------------------\n", + "(351, 416) (416, 408)\n", + "[ 0.99627876 -0.24119866 -0.57631266 0.7955999 -0.8359073 0.47763675\n", + " -0.14629948 0.971156 -0.7675923 0.59845716 -0.45719984 0.86825705\n", + " -0.91944563 -0.50656193 -0.2165582 0.95692414 -0.544702 0.8034633 ] [ 0.287837 -1.0031514 -0.90988153 0.4695353 -0.71728283 -0.7830639\n", + " 0.17799604 0.98139006 -0.7249609 -0.7165208 -0.81639785 0.59817034\n", + " -0.07814442 -1.059892 -0.05741081 -1.0517857 -0.6911336 -0.789837 ]\n", + "348.2289974631019 427.44571449240226\n", + "448.20905622211166 345.5841901247025\n", + "376.33290908276916 371.6651312148954\n", + "410.60295896139723 407.0211313527966\n", + "383.1503325283377 374.76338461857955\n", + "447.2879421014276 365.29746516668723\n", + "343.6537263627124 402.6659952307135\n", + "392.11196818381745 406.6960924104822\n", + "392.77605258788356 387.6782349048581\n", + "--------------------\n", + "(368, 400) (413, 393)\n", + "[ 0.82214755 -0.5640469 -0.5673741 0.8202904 -0.9544657 0.30152568\n", + " 0.13444372 0.9718925 -0.8209629 0.5240966 -0.36654434 0.95175916\n", + " -0.08897487 -1.0588479 0.27960712 0.95052236 -0.42658663 0.8924304 ] [ 0.43972424 -0.94106686 -0.8104618 0.60543185 -0.5586636 -0.8843866\n", + " 0.754384 0.64609724 -0.5000263 -0.8858901 -0.5920873 0.81549025\n", + " 0.04511068 -1.0583081 0.53200316 -0.8617422 -0.19609879 -0.9985816 ]\n", + "353.8362114629093 420.64495639502115\n", + "450.28942108599983 343.08248892519356\n", + "377.32971484492487 370.4672175548997\n", + "425.84885647235114 408.0023411847227\n", + "383.20484514953785 376.1826093842604\n", + "443.3012372262385 370.9997624142467\n", + "327.8761357207859 396.6283962834283\n", + "390.7982421113518 406.70636587596283\n", + "391.4836648651918 388.77467954885054\n", + "--------------------\n", + "(401, 377) (461, 379)\n", + "[ 0.6912531 -0.7316698 -0.50464046 0.8743306 -0.33451954 -1.0013647\n", + " 0.77170813 0.6097189 -0.06105924 -1.021955 -0.14282346 0.98302346\n", + " 0.34576118 -0.98650265 0.9596307 -0.36265165 0.87385845 -0.4607692 ] [ 0.38299266 -0.94597036 -0.89460343 -0.4647112 -0.1268279 -1.0469035\n", + " 0.65194005 -0.74634224 -0.03746397 -1.0055948 -0.32900232 -0.9668143\n", + " 0.16231601 -1.0071468 0.37021172 -0.98215485 0.12627546 -1.0220848 ]\n", + "352.3014150072906 423.00852629631095\n", + "446.3609961654074 350.81883478153344\n", + "376.2545723613689 368.7334522946119\n", + "426.43522533015914 409.19281922611293\n", + "390.5370606834931 376.3748657669917\n", + "438.9287801509877 371.4893283931945\n", + "339.36162720095103 398.6037505417684\n", + "390.15205029890586 405.70530523549036\n", + "395.69094448383134 387.0687349881796\n", + "--------------------\n", + "(411, 404) (347, 389)\n", + "[ 0.36498147 -0.962702 -0.8628784 0.5445778 -0.7242284 -0.7252487\n", + " 0.3398401 0.914428 -0.70188713 -0.7239893 -0.7371 0.6894214\n", + " -0.10117885 -1.0722742 0.17021213 -1.0186558 -0.6728106 -0.78588676] [ 1.0106233 0.03454498 -0.37268308 0.91294193 -0.51706755 0.8224539\n", + " 0.17952383 0.9670186 -0.35176232 0.9307803 -0.18355861 0.99590445\n", + " 0.84824026 -0.5544112 0.39098984 0.84118986 -0.00997769 0.976368 ]\n", + "348.3249302809231 427.76149548191535\n", + "445.963293151174 348.600989057308\n", + "377.056231660226 370.1039848877694\n", + "394.23287985695833 397.7686290286848\n", + "381.91643891879147 375.8043102128182\n", + "441.2833685446322 371.62231200380603\n", + "341.4832761989246 397.44046315528277\n", + "387.6616440561893 407.8997191101123\n", + "392.93074739357536 388.5306144621487\n", + "--------------------\n", + "(349, 399) (406, 387)\n", + "[ 1.0074635 -0.05770258 -0.47274807 0.9131838 -0.69769263 0.66630536\n", + " 0.10711221 0.9710672 -0.5531367 0.7652116 -0.29438806 0.9766563\n", + " -0.21322976 -1.0304537 0.17851464 0.96716714 -0.21934137 0.932771 ] [ 0.5919997 -0.8499522 -0.70674384 0.72322524 -0.5498374 -0.89523065\n", + " 0.71082544 0.67030317 -0.43218404 -0.901219 -0.37226477 0.94546896\n", + " 0.15910415 -1.056289 0.77364504 -0.6497609 0.18564409 -1.0106796 ]\n", + "347.34749839416867 427.8520028138533\n", + "444.10168018876425 349.76660984408284\n", + "377.29639835499773 369.3706587271525\n", + "425.2465644894801 407.4102703568887\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "381.7131561509957 375.35314561615985\n", + "462.127456400837 364.900623905315\n", + "339.4508279973627 397.0240085082334\n", + "389.6240940605903 406.4981827794413\n", + "378.72926483479637 392.0091551476997\n", + "--------------------\n", + "(349, 403) (362, 418)\n", + "[ 0.99521685 -0.0609294 -0.51427954 0.8846368 -0.74191135 0.6199278\n", + " 0.02207924 0.9768278 -0.62614465 0.74574274 -0.35174212 0.97030234\n", + " -0.57856166 -0.8345304 0.06348369 0.9759226 -0.30704188 0.9304039 ] [ 0.61101156 -0.788623 -0.65198785 0.77372426 -1.0043525 0.28675\n", + " -0.1519188 0.9675546 -0.8963344 0.45175916 -0.52348536 0.84053946\n", + " -0.72089714 -0.7456494 -0.39436528 0.8915131 -0.67447656 0.7217253 ]\n", + "347.3885905836952 429.32065578951324\n", + "448.32252866390616 345.25930308540006\n", + "375.25286628234113 371.58134348598395\n", + "443.8756604935079 405.1444746485707\n", + "384.6437383685595 373.0726072478408\n", + "437.7334704096729 370.8334280988345\n", + "340.1103886503355 396.8370272521941\n", + "389.89523563162766 405.6602317468902\n", + "393.9095787381807 388.1794237593947\n", + "--------------------\n", + "(425, 373) (395, 363)\n", + "[ 0.56561846 -0.83679557 -0.7408513 0.68973035 -0.08966307 -1.0371017\n", + " 0.99523026 0.05052018 0.06210874 -1.0343094 -0.02163806 1.0154421\n", + " 0.30625176 -0.98364705 0.6815924 -0.7542648 0.49057847 -0.87614715] [ 0.8199331 -0.58857614 -0.23986256 0.960584 0.3109546 -0.98393995\n", + " 0.79896516 0.57226545 0.7685911 -0.6657093 0.16595282 0.98321265\n", + " 0.5779686 -0.8711672 1.0121899 -0.13429834 1.0049715 -0.05033331]\n", + "352.77979847386916 421.8160782823978\n", + "446.21651719689146 350.21097181107643\n", + "376.59842901508637 368.8154219763985\n", + "426.74197836068663 407.3163775436131\n", + "384.21750011091535 375.4489284252829\n", + "450.9682444703976 372.4466426783439\n", + "340.071353957851 399.4419516882401\n", + "389.405490895114 405.16502569459874\n", + "393.6191089682203 390.57100946609376\n", + "--------------------\n", + "(365, 399) (360, 418)\n", + "[ 0.86326826 -0.50767094 -0.53630894 0.84107506 -0.9246812 0.37654087\n", + " 0.13426471 0.9841378 -0.75759363 0.59535664 -0.34629184 0.9511621\n", + " -0.06147651 -1.0648351 0.27906635 0.94138753 -0.34916106 0.9172329 ] [ 0.671213 -0.72458375 -0.6304931 0.77442116 -0.9616601 0.33209783\n", + " -0.14782563 0.9732089 -0.8649745 0.47645983 -0.53407675 0.8398428\n", + " -0.73827803 -0.71933377 -0.35019138 0.9012777 -0.66392404 0.73655 ]\n", + "346.4388551272584 430.56226972100296\n", + "449.58475224904 345.06478949488803\n", + "375.27627670020206 373.7642812520019\n", + "428.26401813967277 407.63103239949504\n", + "383.2770586814605 375.7423728804962\n", + "423.19475211007466 377.81289924473964\n", + "340.0862859363706 397.5616474615247\n", + "389.9013511607084 406.3817945073214\n", + "392.8321477267974 388.405196661484\n", + "--------------------\n", + "(346, 407) (434, 368)\n", + "[ 1.0090165 0.07905609 -0.49936748 0.8765172 -0.77249086 0.6010014\n", + " -0.02078138 0.9983604 -0.6264757 0.73874974 -0.35863793 0.94371253\n", + " -0.9056034 -0.4538658 -0.01974121 0.9516252 -0.35321936 0.9054778 ] [ 0.55908597 -0.85884964 -0.83854705 0.57765424 0.04193112 -1.0509715\n", + " 0.9631695 -0.12552169 0.15019712 -0.9657592 0.49955088 0.85250837\n", + " 0.32193744 -0.9872306 0.6857119 -0.75993836 0.4944288 -0.8770363 ]\n", + "347.36313351820553 424.39808052541616\n", + "446.62599775716535 349.6715744925264\n", + "374.49512997837877 370.37408204437105\n", + "429.1429986728872 405.26933578418317\n", + "382.5545943883324 376.00091610394077\n", + "439.7530326238707 371.3711487049395\n", + "341.56173071901645 398.1442721569234\n", + "391.83264289496947 406.0492140332524\n", + "407.0969179040138 383.16660049485444\n", + "--------------------\n", + "(418, 374) (444, 400)\n", + "[ 0.6509717 -0.7958972 -0.6606885 0.77966183 -0.11696842 -1.0073625\n", + " 0.9265647 0.25145572 0.04721312 -1.0410763 -0.0478834 1.0140041\n", + " 0.3225586 -0.98742205 0.7289332 -0.6945654 0.50801235 -0.8938633 ] [ 0.28249425 -0.96143526 -1.0060421 0.04792976 -0.39551365 -0.9605394\n", + " 0.5214947 -0.8666016 -0.35351104 -0.9458065 -1.0289899 -0.09642851\n", + " 0.02656576 -1.0157285 0.12578213 -1.0464978 -0.2044529 -1.0088719 ]\n", + "353.8127803038824 426.4993250046738\n", + "446.3846445316744 349.94673902459436\n", + "366.2688965100474 367.99331899250654\n", + "427.7154580880402 409.7995457325182\n", + "379.151271009654 375.76180146138825\n", + "441.45969420870006 372.8921839593498\n", + "329.2184146022121 403.00205034443496\n", + "386.6587658200355 406.8920291794022\n", + "391.11116229393895 389.2818238772395\n", + "--------------------\n", + "(447, 399) (414, 364)\n", + "[ 0.26692727 -0.97173387 -1.0159962 -0.02685678 -0.34622374 -0.9691926\n", + " 0.47324732 -0.897725 -0.3281184 -0.95409954 -1.0029947 -0.23907788\n", + " 0.02893981 -1.0077622 0.12426385 -1.0462649 -0.17663531 -0.99271864] [ 0.7214154 -0.72175574 -0.39042225 0.92081326 0.12652826 -1.0168979\n", + " 0.9320799 0.27030534 0.3650794 -0.9537779 0.27313817 0.9648412\n", + " 0.4490328 -0.9516827 0.85866433 -0.49520794 0.7707365 -0.63729846]\n", + "353.20730950189545 424.7640780402657\n", + "445.71930477354476 350.5511017513598\n", + "365.80891856766607 369.99618816242366\n", + "427.1800829833378 409.4483252475899\n", + "381.4528670031225 376.4580939996721\n", + "440.4414965412788 371.48535045265345\n", + "332.87400686589535 402.2773454240479\n", + "389.90399710194833 405.78123600698035\n", + "393.0037745542809 389.39240359013047\n", + "--------------------\n", + "(419, 368) (422, 390)\n", + "[ 0.6801772 -0.75755674 -0.5384141 0.85513645 0.03210551 -1.0361831\n", + " 0.9708131 0.16579649 0.20413007 -0.99664855 0.19388266 0.9868856\n", + " 0.3847026 -0.97835565 0.79862463 -0.6264526 0.6345575 -0.78753555] [ 0.45583713 -0.92292345 -0.8756455 0.52233654 -0.41486406 -0.9865953\n", + " 0.9648833 0.24967678 -0.3434764 -0.9412325 -0.6800806 0.7421178\n", + " 0.10756972 -1.0378529 0.4752909 -0.9057373 -0.01256965 -0.9855812 ]\n", + "360.8698489568442 420.1925300164815\n", + "444.821346374167 351.7422698610089\n", + "373.0658817888743 369.4232410274193\n", + "424.22693436727485 398.6060539443789\n", + "382.3073725746917 375.5152555738368\n", + "441.2392038878138 372.36909403429985\n", + "341.9506964357952 398.2968220209592\n", + "387.571038886617 408.06678495091006\n", + "392.1684710707053 389.61952817123876\n", + "--------------------\n", + "(326, 421) (346, 418)\n", + "[ 0.31508318 0.83468306 -0.51253086 0.9005211 -0.69965523 0.6805604\n", + " -0.14625634 0.9986315 -0.5964035 0.7641348 -0.42244685 0.87054497\n", + " -0.7645145 0.4627595 -0.23295642 0.9788597 -0.41984951 0.8079175 ] [ 0.9703537 0.24290606 -0.60600585 0.8005912 -0.7984998 0.5247744\n", + " -0.12587649 0.9636625 -0.7299643 0.6398268 -0.43325648 0.8868897\n", + " -0.985848 -0.26158962 -0.23341045 0.9215392 -0.49766865 0.82069594]\n", + "348.91649135600005 429.6507100444274\n", + "390.6381157084195 384.21126859936936\n", + "381.58140575601544 363.8591181680238\n", + "350.4755335683867 417.4153923797431\n", + "380.99028214009246 378.08035607249286\n", + "2412.2916574458422 -591.4087536575171\n", + "340.4580470334846 397.11418733659343\n", + "461.04862479291023 388.86010868993304\n", + "431.24635911489895 366.30674809689333\n", + "--------------------\n", + "(370, 406) (366, 399)\n", + "[ 0.73237383 -0.7154301 -0.63477033 0.7869808 -0.9977008 0.18951902\n", + " 0.03648258 0.9774137 -0.913388 0.3698312 -0.42862555 0.89110345\n", + " -0.31654558 -0.9915433 0.05163531 0.9546746 -0.58693415 0.7952511 ] [ 0.8481299 -0.52807873 -0.54729176 0.82707876 -0.93853575 0.3544741\n", + " 0.13855338 0.98263377 -0.7735547 0.58322567 -0.35031557 0.95083636\n", + " -0.05928896 -1.0713582 0.29461023 0.94950694 -0.3604676 0.9130242 ]\n", + "346.94945230783196 429.596460393803\n", + "436.58857034902826 352.29037015435614\n", + "376.72249318814715 370.6102159403238\n", + "432.0771523044714 408.31706896361413\n", + "380.7620314601917 379.4205483140941\n", + "445.2707141101778 369.79438070971304\n", + "344.3141233047836 397.7999035965509\n", + "392.4790825532782 407.2158219814645\n", + "394.99463583829515 387.5527374894512\n", + "--------------------\n", + "(374, 366) (432, 390)\n", + "[ 0.9502933 -0.41185978 -0.20718412 0.9641055 0.60569036 0.7229293\n", + " 0.60419095 0.74774474 0.73342925 0.626246 0.07230781 1.0073382\n", + " 0.72900003 -0.7436858 0.9261831 0.36757648 0.77578837 0.6342085 ] [ 0.42280805 -0.94132626 -0.975494 0.3320136 -0.31812882 -0.97023296\n", + " 0.9688129 -0.22927785 -0.27766624 -0.99016184 -0.86968505 0.4998749\n", + " 0.11199932 -1.0467407 0.36280727 -0.9443363 -0.0148611 -1.0245129 ]\n", + "347.06400727005337 428.15001916162\n", + "445.39002777422115 350.658441348095\n", + "383.7706463045334 374.18612032639953\n", + "427.45743374596555 409.1945492219181\n", + "382.68431122646945 376.1706490646647\n", + "442.9499882502869 370.9493039130226\n", + "339.4099639284288 399.9069780114991\n", + "389.93831985213967 406.1598120688112\n", + "393.1594781961711 389.4366143231807\n", + "--------------------\n", + "(392, 378) (438, 406)\n", + "[ 0.79509145 -0.68382776 -0.4429266 0.8738936 -0.574428 -0.8826356\n", + " 0.6100827 0.7398644 -0.23569979 -1.0060185 -0.12177009 0.98833954\n", + " 0.38846815 -1.0014173 0.9778896 -0.01879266 0.9709406 0.15761909] [ 0.20941818 -1.0240971 -1.0215565 0.13855545 -0.45933503 -0.8901301\n", + " 0.35533628 -0.99191624 -0.43835905 -0.8755212 -1.0760236 0.08653088\n", + " -0.07340018 -1.0814207 -0.00793446 -1.0431666 -0.3879283 -0.92086136]\n", + "352.9623198277364 423.38939152248463\n", + "445.473690617388 350.8972504747955\n", + "429.6039371725488 409.00772205025197\n", + "373.34891561626887 373.6302429088081\n", + "440.73440573623384 371.99559258156734\n", + "337.41858345272294 399.17313340769067\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "391.4687091953977 405.64609790495217\n", + "393.5023657228133 387.25463920132125\n", + "--------------------\n", + "(444, 373) (415, 356)\n", + "[ 0.46435362 -0.8817472 -0.9895897 0.08380553 -0.05736175 -1.0684743\n", + " 0.8834197 -0.42591512 0.0828855 -1.0401559 0.05679608 -0.582279\n", + " 0.25411344 -1.039172 0.5462575 -0.8931146 0.3149686 -0.959223 ] [ 0.7627771 -0.7039611 -0.18094037 0.9909841 0.31943217 -0.9643321\n", + " 0.9713284 0.19738103 0.54102856 -0.869782 0.5175326 0.85835505\n", + " 0.52154034 -0.9065438 0.89247084 -0.46573794 0.8275412 -0.5665743 ]\n", + "357.0524041924426 418.7891223450368\n", + "445.91775418539845 350.3548388106337\n", + "374.88115222480246 369.28930890929365\n", + "426.02903703502164 410.27480348585397\n", + "379.3935935148811 378.14820321220753\n", + "443.30758898277827 373.06753846355866\n", + "342.16579248138214 397.9019789613044\n", + "388.3735882153678 407.02289304353013\n", + "391.5754698004391 390.21401700089893\n", + "--------------------\n", + "(360, 372) (354, 391)\n", + "[ 1.0181619 -0.23689088 -0.28527492 0.93248606 -0.12970354 0.92988634\n", + " 0.49332374 0.83885235 0.15257381 0.9553039 -0.05874961 1.0035915\n", + " 0.81366783 -0.64007246 0.8087248 0.6049297 0.46788415 0.8365675 ] [ 1.0203004 -0.17832401 -0.42071322 0.89466935 -0.66434544 0.73124075\n", + " 0.22210473 0.9619185 -0.44592023 0.8563437 -0.25063032 0.96857065\n", + " 0.45542517 -0.9599171 0.3929343 0.8818277 0.00414806 0.94079256]\n", + "349.231683521185 418.28244615923984\n", + "458.4605266528233 341.8780279980228\n", + "377.61799435203557 369.5425855485673\n", + "417.070587898021 405.56285108890694\n", + "383.33157404105214 375.7263398746673\n", + "447.139522281831 366.89890748300564\n", + "339.7264441822356 397.7719878730412\n", + "384.31679574786574 404.50889500008606\n", + "394.28923471023836 391.17763799523635\n", + "--------------------\n", + "(397, 356) (453, 364)\n", + "[ 0.8527907 -0.557952 -0.10979348 0.97475064 0.58913386 -0.8531483\n", + " 0.8280913 0.45922187 0.82272863 -0.5403615 0.32220066 0.93511313\n", + " 0.6421354 -0.8442225 0.98367363 -0.14892344 1.0091193 -0.11898871] [ 0.53146917 -0.8359908 -0.9072906 -0.4474815 0.09118942 -1.0082682\n", + " 0.8518292 -0.49223277 0.2129334 -0.9506903 0.6277577 -0.74072367\n", + " 0.33712897 -0.9936897 0.56143767 -0.8423133 0.39499283 -0.9555441 ]\n", + "348.15776898092463 430.65194742574613\n", + "446.31462194358573 350.44532395742476\n", + "375.22904078117006 371.0337402689384\n", + "426.68776040768006 409.5344187922286\n", + "381.1805387229632 380.085955786793\n", + "443.5245559971911 372.03040460747866\n", + "332.92290040413593 404.738540271732\n", + "389.3676085016807 406.413704766837\n", + "393.1389553514763 388.7447417420703\n", + "--------------------\n", + "(366, 387) (421, 373)\n", + "[ 0.93345547 -0.43100795 -0.42365602 0.9201608 -0.79622096 0.53110015\n", + " 0.3031103 0.9081392 -0.4921364 0.8247863 -0.20597453 0.97819346\n", + " 0.40771398 -0.98358923 0.6220246 0.71046597 0.10670075 0.98502773] [ 0.6011975 -0.81162125 -0.67313707 0.75832194 -0.08527112 -1.0206538\n", + " 0.98187363 0.14088352 0.0604135 -1.0337557 -0.02673378 1.0235758\n", + " 0.32888716 -0.9789662 0.7163331 -0.7191606 0.51426196 -0.88328236]\n", + "347.2349101241392 427.64049273007345\n", + "447.5015844366827 349.4754293841173\n", + "377.748645856647 369.38652758586494\n", + "425.87630426212854 406.98495852453965\n", + "386.0388486135937 375.0431497376553\n", + "434.11398850419073 372.6574931681838\n", + "309.00817173127257 410.6240539795827\n", + "387.7910283893476 406.07840262142304\n", + "392.0982178208119 389.82702627049775\n", + "--------------------\n", + "(446, 366) (396, 353)\n", + "[ 0.53396016 -0.8324322 -1.0313967 0.0170755 0.05506793 -1.0163999\n", + " 0.8893197 -0.38706148 0.15482977 -1.024455 0.8828809 -0.45287415\n", + " 0.31653768 -1.0021026 0.5840854 -0.8312953 0.4443487 -0.91737765] [ 0.89111865 -0.5477663 -0.05362016 0.9778902 0.6732264 -0.79363656\n", + " 0.8519439 0.4717734 0.88138866 -0.4580459 0.36733127 0.93495977\n", + " 0.6618109 -0.8247171 1.0103873 -0.10542271 1.0148821 -0.09004667]\n", + "350.25882186176716 427.41278015416566\n", + "446.2608505115346 350.2440738213556\n", + "376.21768707631963 369.780763947886\n", + "427.16423067719154 409.2774179045199\n", + "384.4063493625017 375.30888268123\n", + "443.16348421097683 371.5298049798338\n", + "336.82639760234133 400.485049189999\n", + "390.5807848689469 404.93869653527594\n", + "392.5494431705098 391.88975844337074\n", + "--------------------\n", + "(455, 389) (417, 343)\n", + "[ 0.3263714 -0.95503527 -1.015124 -0.2452295 -0.21824971 -1.0274074\n", + " 0.58758026 -0.8031604 -0.16643684 -0.9946489 -0.8063546 -0.6645771\n", + " 0.10295977 -1.035613 0.2634286 -0.9985683 -0.02132205 -1.0333943 ] [ 0.785731 -0.67437434 0.25767234 0.94148713 0.5347662 -0.8416966\n", + " 0.97307724 0.11311401 0.7061378 -0.7438129 0.7394324 0.6722451\n", + " 0.5987042 -0.84312195 0.9132712 -0.40351424 0.8584665 -0.49444175]\n", + "345.3618390117351 426.46747472846897\n", + "445.79106094592885 350.8797296820235\n", + "372.26175375268554 371.42411088525574\n", + "424.906423843635 411.0160141434688\n", + "381.5007142850659 376.70119913267655\n", + "417.94119956246897 344.0352674258799\n", + "335.48851318616744 400.88173198833124\n", + "388.980497633057 406.41636017953874\n", + "391.26369508126663 387.684927051812\n", + "--------------------\n", + "(412, 346) (381, 366)\n", + "[ 0.7920807 -0.6024333 0.12193232 0.9652549 0.559476 -0.8506567\n", + " 0.97111356 0.20510174 0.7325164 -0.6992927 0.64300734 0.74929696\n", + " 0.611977 -0.8269932 0.95613843 -0.33258745 0.9220803 -0.40277773] [ 0.9192 -0.47185236 -0.27316964 0.95507663 0.7250058 -0.7135342\n", + " 0.6633199 0.71290606 0.9613648 0.23088281 0.12172715 1.0399212\n", + " 0.6437301 -0.80771965 0.92174196 0.16748683 0.92296034 0.47597426]\n", + "348.21928299967533 429.8590305736231\n", + "439.000540289283 349.41074521608306\n", + "379.9154833675707 367.1019526393024\n", + "424.83895011485254 406.78972467417515\n", + "383.39339758974063 375.9657141062655\n", + "443.88354480445696 373.36078405088955\n", + "329.39238820144897 407.1298309439569\n", + "389.2500038145682 411.4027858615503\n", + "393.0538300754355 389.373525165072\n", + "--------------------\n", + "(410, 404) (440, 393)\n", + "[ 0.36956602 -0.9590116 -0.85242754 0.55596817 -0.73072815 -0.7106234\n", + " 0.32071897 0.9131319 -0.7217864 -0.71076363 -0.7159673 0.70351315\n", + " -0.09779593 -1.0705385 0.1770841 -1.014095 -0.6968589 -0.76035655] [ 0.3340157 -0.933084 -0.9946059 0.12989159 -0.34406516 -0.98818934\n", + " 0.78073317 -0.6238191 -0.2689816 -0.9824271 -1.04049 0.09844709\n", + " 0.06340905 -1.0413139 0.2647513 -1.0084773 -0.07889257 -1.0303406 ]\n", + "445.7147456037832 349.2410504155013\n", + "378.46807758971215 371.57598512366343\n", + "426.56272377628784 409.817319120293\n", + "384.0954791425135 377.693743545561\n", + "442.0448437340966 371.3878735203367\n", + "349.7472721579279 398.4957874493567\n", + "374.4582231751792 410.2064048828665\n", + "394.168631073303 389.49071542193957\n", + "--------------------\n", + "(400, 399) (369, 416)\n", + "[ 0.48883507 -0.94229907 -0.7403478 0.6809996 -0.8351553 -0.5944649\n", + " 0.36937568 0.93053126 -0.85067743 -0.61295986 -0.563806 0.8452144\n", + " -0.03718828 -1.0679247 0.6459723 -0.81374156 -0.8842809 -0.5333344 ] [ 0.47603777 -0.88158387 -0.7017827 0.786889 -1.0244831 0.14984283\n", + " -0.1265359 0.9666313 -0.9626901 0.31883907 -0.54124177 0.86907554\n", + " -0.5314478 -0.8929571 -0.39972737 0.87490124 -0.77508557 0.660745 ]\n", + "454.5159030865937 339.73310596282414\n", + "376.3467558656559 365.7698913358483\n", + "424.51800852251245 408.732457796024\n", + "382.61924490556686 374.87865615457173\n", + "452.08323930168467 364.25752441029425\n", + "336.7343887071311 396.79690506305764\n", + "391.5813708971578 405.68295843882515\n", + "393.1600153111415 387.659144971993\n", + "--------------------\n", + "(410, 401) (356, 420)\n", + "[ 0.3879587 -0.9587689 -0.83864933 0.56384975 -0.70303583 -0.7459767\n", + " 0.43859494 0.85191995 -0.68902636 -0.7356768 -0.6883148 0.74703175\n", + " -0.06406062 -1.0953227 0.30900544 -0.97340846 -0.60645044 -0.8354218 ] [ 0.76943785 -0.65154904 -0.6352543 0.78269434 -0.9008005 0.36699313\n", + " -0.16708925 0.9681819 -0.84761983 0.5182454 -0.5357365 0.8278088\n", + " -0.8463992 -0.6002686 -0.34018287 0.8898782 -0.6878847 0.7535814 ]\n", + "352.3278083066798 424.3366241660095\n", + "446.74167223006555 346.35178863339956\n", + "376.5746232072985 369.49869591099156\n", + "424.0827784409721 408.2502532016223\n", + "383.0497049602371 375.7586672723031\n", + "468.1543584583797 347.41659201078363\n", + "339.60493648034753 396.88290171319\n", + "384.65607199438 409.04534957147285\n", + "391.5148959738327 387.5812702428701\n", + "--------------------\n", + "(388, 381) (373, 401)\n", + "[ 0.8086949 -0.66984403 -0.4592292 0.8687578 -0.83795285 -0.7149666\n", + " 0.5603183 0.77678156 -0.8323521 -0.67669237 -0.13920887 1.0070289\n", + " 0.3656206 -0.9880181 0.9571413 0.09819416 0.7883181 0.577916 ] [ 0.73081625 -0.7164978 -0.5986346 0.8101677 -0.9811939 0.14181356\n", + " 0.13508661 0.9565693 -0.90609396 0.35981134 -0.39112014 0.9123606\n", + " -0.10362841 -1.0181813 0.2987889 0.90872115 -0.514695 0.84939045]\n", + "362.905877992321 411.2958423135703\n", + "430.3992657793656 358.58756076818827\n", + "377.644754958433 368.86347924904584\n", + "426.127552702751 408.5026668493131\n", + "383.2581660708098 375.16740058682086\n", + "434.7188916682062 374.54171095332384\n", + "342.3764412524374 397.88320656743116\n", + "390.6470946682321 406.8023855964875\n", + "393.5382966931983 388.55462592448765\n", + "--------------------\n", + "(454, 395) (370, 381)\n", + "[ 0.3039993 -0.9678278 -1.0071951 -0.19830017 -0.29941687 -1.0011524\n", + " 0.4846526 -0.8865229 -0.2641444 -0.9641388 -0.90218395 -0.5090232\n", + " 0.05648509 -1.0362732 0.19092035 -1.0057191 -0.10187915 -0.99487805] [ 0.91868716 -0.45767638 -0.3916132 0.92854506 -0.80936605 0.53467965\n", + " 0.41648972 0.86500055 -0.31195098 0.91639775 -0.15890276 0.9948184\n", + " 0.5073093 -0.956664 0.7837063 0.5968023 0.33347926 0.9177254 ]\n", + "346.14860568717677 428.8766338269916\n", + "445.014740313404 349.3625732802497\n", + "376.1353109750387 371.7128343498734\n", + "428.27958380238294 409.06107755418043\n", + "384.6707696686683 376.0059228266387\n", + "439.8096500922994 369.84926800306863\n", + "330.95109696270174 401.7071394296841\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "389.924109562412 407.16382547365555\n", + "390.68442127389426 388.51625299514563\n", + "--------------------\n", + "(405, 401) (337, 417)\n", + "[ 0.41982478 -0.9387523 -0.7968022 0.6207836 -0.75540084 -0.6770783\n", + " 0.3662357 0.90313226 -0.77424765 -0.65810627 -0.63165057 0.7901159\n", + " -0.0474645 -1.0627432 0.38079828 -0.95174134 -0.7179375 -0.7034494 ] [ 0.6830476 0.67886114 -0.5328975 0.8466488 -0.77315825 0.6367198\n", + " -0.10993523 0.99224657 -0.6459534 0.72961295 -0.42762962 0.91198355\n", + " -0.96314645 0.17595704 -0.20456979 0.93916667 -0.42411038 0.8868443 ]\n", + "346.9152799735931 426.9764007421109\n", + "445.9717268359422 348.4110436168123\n", + "376.42814587047 369.12303318442866\n", + "421.3970795312715 407.6492982001745\n", + "383.5611620552022 375.7776757020267\n", + "453.05847648551963 362.58011261550524\n", + "340.44977720874704 398.11704210701066\n", + "398.4816677884925 403.6080298011358\n", + "393.9785070647811 389.7515106857844\n", + "--------------------\n", + "(452, 397) (402, 347)\n", + "[ 0.2965304 -0.98958695 -1.018073 -0.13577452 -0.30571678 -0.9989333\n", + " 0.4694072 -0.88831925 -0.29523447 -0.97582793 -0.9560378 -0.43144348\n", + " 0.01454686 -1.0409017 0.16101758 -1.0347363 -0.15159443 -0.99613184] [ 0.844907 -0.5487248 0.07267626 0.9904359 0.67906857 -0.7774259\n", + " 0.9352113 0.3448613 0.8254093 -0.54113495 0.5311759 0.84948117\n", + " 0.6770586 -0.7909259 1.010906 -0.20142326 0.97788924 -0.25593892]\n", + "349.59960616429225 427.6843472137139\n", + "445.7600259383768 350.2110368487012\n", + "372.5832831720097 372.6950502983992\n", + "425.5848246656137 410.9583530242321\n", + "382.92176079921524 376.1005395336149\n", + "440.22133216240616 370.899589566405\n", + "341.7906500912104 398.5402034160045\n", + "390.1188207970454 406.6294664014675\n", + "391.33018877202494 387.7670803016149\n", + "--------------------\n", + "(424, 373) (333, 415)\n", + "[ 0.5721383 -0.83109844 -0.71888274 0.7089012 -0.08639133 -1.0364693\n", + " 0.99312097 0.07482152 0.06189832 -1.0379896 -0.02873268 1.0165162\n", + " 0.31758806 -0.98418653 0.6912468 -0.7452023 0.5006644 -0.883048 ] [ 0.6708892 0.72343916 -0.51550967 0.8575102 -0.7086228 0.6933856\n", + " -0.07142532 0.9971429 -0.60285336 0.76684475 -0.39889267 0.9387343\n", + " -0.8761896 0.41778004 -0.1473022 0.965963 -0.39113688 0.8921561 ]\n", + "345.77745344121115 426.84935778234967\n", + "454.7730374890553 341.79366938869833\n", + "377.86001490933995 369.1541604271659\n", + "426.65883375724565 408.2912337073417\n", + "383.3409828101949 375.42461462707877\n", + "432.39907289106486 372.7625931625237\n", + "340.1201034974497 400.0672810598261\n", + "387.7168802800713 406.6560814345108\n", + "407.633487714832 382.27937160359284\n", + "--------------------\n", + "(381, 387) (419, 361)\n", + "[ 0.782884 -0.64884764 -0.50054103 0.8700788 -0.98431593 -0.2270387\n", + " 0.40797773 0.89309865 -0.960934 0.20059425 -0.23551175 0.98463756\n", + " 0.2601758 -1.0515006 0.91937673 0.4200486 0.16208518 0.96308976] [ 0.7256059 -0.73848546 -0.36459994 0.93793786 0.20242122 -1.0094169\n", + " 0.97902566 0.15976194 0.39868242 -0.9421235 0.44469598 0.9095277\n", + " 0.46359795 -0.9364603 0.83924866 -0.55080223 0.7354826 -0.67823124]\n", + "330.3909372467104 448.06368026971194\n", + "441.1876482414074 352.375105386109\n", + "376.94801786090477 369.4328159015758\n", + "426.6454273000604 407.8513567817655\n", + "383.27130191558217 376.11947266790696\n", + "442.2254996280233 372.3557025651284\n", + "351.97185506385836 394.182516603655\n", + "389.59277980189904 405.8073519361223\n", + "393.13978960101804 389.04309087858775\n", + "--------------------\n", + "(417, 352) (337, 424)\n", + "[ 0.7598778 -0.6927547 -0.0585933 0.99402946 0.39674684 -0.9323361\n", + " 0.99464136 0.16295491 0.59846985 -0.82579154 0.61934406 0.76997036\n", + " 0.5552691 -0.9025973 0.9120249 -0.45720163 0.8468333 -0.5557201 ] [ 0.36428693 0.91865355 -0.5295157 0.84834063 -0.8119498 0.60068524\n", + " -0.17378724 0.98204225 -0.6924152 0.6573547 -0.44403934 0.91157407\n", + " -0.96703714 0.11784301 -0.31504902 0.94303554 -0.4974772 0.83173555]\n", + "347.5471062511042 428.1823958713806\n", + "456.0381888718075 349.6988846955951\n", + "377.98266463390064 368.60345900602107\n", + "426.20954033674235 408.213033898893\n", + "379.6708140529189 379.05330843028406\n", + "442.5759341110651 372.5726141745144\n", + "340.0015452550675 399.36870480953905\n", + "389.7385262089323 406.38113771481164\n", + "390.91197525754745 391.75419878292337\n", + "--------------------\n", + "(431, 355) (437, 375)\n", + "[ 0.6520084 -0.7623158 -0.26026118 0.95556915 0.2808653 -1.015783\n", + " 0.98267573 -0.06266367 0.38950947 -0.93027234 0.80604005 0.54537624\n", + " 0.46210077 -0.9452528 0.8298014 -0.6469255 0.6913179 -0.7595861 ] [ 0.5238802 -0.8738452 -0.9524501 0.32959703 -0.05062805 -1.026785\n", + " 0.9425801 -0.26765487 0.03599175 -1.0086644 -0.4599451 0.9288722\n", + " 0.23995298 -1.0348055 0.62317383 -0.90910506 0.3432994 -0.9396691 ]\n", + "338.7476036765961 433.9034351143214\n", + "445.26561144163543 351.1145827112429\n", + "370.5222946747449 371.7221649549333\n", + "427.6176522774786 408.0411118947723\n", + "378.2247581415754 377.0972462749812\n", + "442.64197325532854 372.206280698332\n", + "347.7597453280964 395.6932253586303\n", + "390.62371248956583 406.79004311320347\n", + "390.26448809724735 392.07438556747115\n", + "--------------------\n", + "(404, 368) (385, 387)\n", + "[ 0.7498627 -0.6900482 -0.38882136 0.9105665 0.05259248 -1.0514138\n", + " 0.834997 0.50071055 0.35681596 -0.9579086 0.10131827 0.994902\n", + " 0.46148506 -0.9412265 0.9389262 -0.35438985 0.92193574 -0.44497553] [ 0.7559296 -0.663607 -0.5243492 0.8406371 -0.94583815 -0.42639896\n", + " 0.44474116 0.8609061 -0.9808665 -0.10831385 -0.25180954 0.97819036\n", + " 0.24722493 -1.0556946 0.9864815 0.2377911 0.24964103 0.9652113 ]\n", + "353.59413755014737 422.77512205416815\n", + "440.3354842405506 352.48437072076774\n", + "377.0421144410275 369.348453038444\n", + "429.0344443863866 409.7480436865177\n", + "383.735428804816 375.548447191389\n", + "443.2721664769356 371.99937680546526\n", + "347.18849560370904 395.85478347278735\n", + "389.6100684335261 406.1249175064116\n", + "393.73869858630064 389.26018107981486\n", + "--------------------\n", + "(390, 366) (430, 388)\n", + "[ 0.84829754 -0.55729204 -0.26349017 0.9674383 0.21348715 -1.0002841\n", + " 0.7584332 0.66646904 0.84622 -0.5057274 0.09474298 0.9817455\n", + " 0.5649567 -0.8551991 1.007237 -0.02023928 0.97890455 0.14078695] [ 0.45450714 -0.9297969 -0.9515411 0.39515465 -0.30497333 -0.9817417\n", + " 0.99768054 -0.14316492 -0.24354921 -1.0073346 -0.7983857 0.62945294\n", + " 0.14037564 -1.0676905 0.41830206 -0.9158759 0.03528253 -1.0190074 ]\n", + "349.78823027142255 427.20945987008946\n", + "445.40236604361394 350.91068689882394\n", + "371.73114029291355 369.89905908549304\n", + "427.0987677319022 408.21792106064123\n", + "383.5620573508148 376.77243584416385\n", + "443.2902802562503 371.1427584084065\n", + "338.48406963559216 400.0321580834499\n", + "389.18334654792886 406.64193600011276\n", + "393.3465820152991 389.26909193524574\n", + "--------------------\n", + "(353, 389) (442, 368)\n", + "[ 1.0263857 -0.1460227 -0.3956343 0.9230983 -0.6014827 0.77170414\n", + " 0.23677276 0.9712636 -0.37021422 0.8877426 -0.22157247 0.9795817\n", + " 0.61490244 -0.8508451 0.4262094 0.85704833 0.05360727 0.96085036] [ 0.5110611 -0.84972113 -0.9838111 0.25162184 0.0431742 -1.0193607\n", + " 0.9129968 -0.3368633 0.09791037 -0.9784302 1.0100894 0.07711297\n", + " 0.29421595 -0.98873806 0.6326968 -0.80529195 0.44046032 -0.90645134]\n", + "347.9391402567885 424.57250721736057\n", + "446.92487467363287 348.7443749044581\n", + "376.37708387123257 370.77940176417314\n", + "427.54668772152763 407.17284730609043\n", + "391.15583956785576 373.0879154601664\n", + "442.0651677908679 368.854269546627\n", + "340.1016189684466 398.3216093819912\n", + "391.1340618454723 407.96403617982804\n", + "394.06726933522714 391.29120425923765\n", + "--------------------\n", + "(397, 346) (376, 357)\n", + "[ 0.74168676 -0.3563437 0.1045808 0.847832 0.7016347 -0.58881366\n", + " 0.7891799 0.22232284 0.791087 -0.3823149 0.37101167 0.68859124\n", + " 0.7786127 -0.7065757 0.89940906 -0.12335167 0.8610653 -0.2511515 ] [ 0.97950125 -0.3720308 -0.11448474 0.9796714 0.97200507 0.20700611\n", + " 0.6433545 0.66530013 0.9404332 0.38241157 0.17417566 0.97428787\n", + " 0.80311203 -0.70687044 0.93712157 0.21365483 0.9615288 0.39844584]\n", + "316.6875934779827 513.1606640307684\n", + "432.57674469942776 350.38842181757246\n", + "378.38210639891844 368.1852196142687\n", + "409.1219734199077 389.0293986375493\n", + "383.16657336865836 374.62416281458394\n", + "437.98101612827793 368.08049513821317\n", + "20.99475950034946 760.3398110366799\n", + "388.17033162446467 410.380839476141\n", + "386.44185268529424 382.1982871275444\n", + "--------------------\n", + "(393, 388) (384, 377)\n", + "[ 0.7100778 -0.7417195 -0.566491 0.84755355 -0.8293664 -0.7593025\n", + " 0.52746576 0.8179253 -0.7715462 -0.6494994 -0.30947077 0.9811861\n", + " 0.1337751 -1.0531293 0.98923767 -0.09571932 0.5276067 0.4690649 ] [ 0.88519347 -0.5897987 -0.43487114 0.9235608 -0.88284916 -0.6953331\n", + " 0.5920275 0.7793327 -0.3228716 0.02067543 -0.07982981 1.0128505\n", + " 0.4300594 -0.94956213 1.0026723 0.22104605 0.763869 0.60581815]\n", + "347.90787665585253 431.16850084088753\n", + "470.1454121217724 336.43726263257673\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "429.27093205601534 411.39049197810897\n", + "442.49260368344585 372.38979687646395\n", + "346.73683129456185 393.8766384204761\n", + "390.9944389213214 408.7270248766072\n", + "390.44277805798663 385.12362295703167\n", + "--------------------\n", + "(408, 344) (386, 389)\n", + "[ 0.856148 -0.62274927 0.19631863 0.95629627 0.6281357 -0.85028327\n", + " 0.8975601 0.288155 0.8034678 -0.5462303 0.6240357 0.72052455\n", + " 0.6328677 -0.80493414 0.97713923 -0.30474874 1.0040989 -0.30309317] [ 0.7365623 -0.722069 -0.5365986 0.8363609 -0.9171241 -0.43045488\n", + " 0.41513607 0.86123514 -0.9838725 -0.19039795 -0.26576388 0.9542603\n", + " 0.1872209 -1.0404876 1.0008922 0.20874356 0.05793056 1.0041155 ]\n", + "344.4042474150225 431.4306573952846\n", + "444.46926986541007 351.4867980147464\n", + "375.9810181992707 367.6536056333552\n", + "429.1196972883326 409.78473195591886\n", + "384.09599956559646 379.1611673383474\n", + "441.96239569253976 373.4143325499351\n", + "340.30833941988635 397.2215792921682\n", + "389.19205680838644 404.3053472795182\n", + "394.27241519906613 389.47727961370083\n", + "--------------------\n", + "(367, 395) (452, 385)\n", + "[ 0.84686315 -0.5232438 -0.5116758 0.8589699 -0.9130852 0.37058604\n", + " 0.21168484 0.93673134 -0.71377975 0.63316005 -0.31918022 0.9556831\n", + " 0.108248 -1.0511813 0.44631907 0.872649 -0.20757662 0.9439014 ] [ 0.37356567 -0.92642283 -0.9990888 -0.14942537 -0.20508188 -1.0229018\n", + " 0.7028664 -0.72062397 -0.09213203 -0.9926882 -0.8156617 -0.65816915\n", + " 0.15965971 -1.0371873 0.31806543 -0.983759 0.0515654 -1.0323277 ]\n", + "347.0247852885084 427.3296206335037\n", + "446.4199592499026 347.69069651973234\n", + "377.14930949105735 369.9931625714884\n", + "427.68665724012 408.71411989547073\n", + "381.6613609943134 378.47182114989374\n", + "440.31198526371236 370.5151726663316\n", + "387.9421810555875 405.7109444101668\n", + "400.8556802389537 387.55468075753356\n", + "--------------------\n", + "(354, 383) (446, 357)\n", + "[ 1.0081877 -0.12470067 -0.35386512 0.9485828 -0.45342994 0.83533114\n", + " 0.3050569 0.92576134 -0.22787805 0.95695835 -0.15071653 0.9975028\n", + " 0.7663095 -0.70236605 0.54005694 0.78708696 0.18843147 0.97647274] [ 0.6151116 -0.8020096 -1.0328503 0.14240195 0.18382603 -0.992528\n", + " 0.9101715 -0.36030358 0.28797236 -0.93581325 0.91803503 -0.19214796\n", + " 0.39908722 -0.964079 0.6646995 -0.7658187 0.5235671 -0.8524207 ]\n", + "347.91073935142526 432.2308276275328\n", + "447.2093673418992 348.22860827869334\n", + "379.0575243645976 369.39840989464426\n", + "426.27914180371295 406.8174253170639\n", + "387.2001900141463 375.0941234827235\n", + "443.3849087417316 369.4944907600572\n", + "336.1528067335166 402.47200224080996\n", + "388.6516298163489 406.77609387080054\n", + "391.79497048914413 390.2933543968332\n", + "--------------------\n", + "(409, 393) (381, 363)\n", + "[ 0.48423362 -0.9440393 -0.79221123 0.6629481 -0.59223086 -0.85295105\n", + " 0.6722057 0.7363185 -0.56186736 -0.8475925 -0.53172946 0.8593018\n", + " 0.05114754 -1.0447689 0.5914439 -0.83192605 -0.25159267 -1.0085043 ] [ 0.9554453 -0.46654522 -0.23430409 0.9645057 0.8913492 -0.46916804\n", + " 0.6682732 0.7195135 1.0087132 0.1953872 0.16079201 0.98394275\n", + " 0.6574619 -0.7937688 0.9036528 0.14419588 0.95192814 0.43946907]\n", + "352.0990499782109 422.1866593054095\n", + "447.65523678797496 346.8076754155121\n", + "376.92984808455253 370.73268061521384\n", + "660.877785808513 622.9462381356094\n", + "383.54212826047194 376.1240282834005\n", + "441.50329268172356 372.8872075763795\n", + "340.7460816015764 396.3414278091469\n", + "388.1521444515115 407.82143400278323\n", + "393.0080592872348 389.0104731661181\n", + "--------------------\n", + "(402, 377) (418, 375)\n", + "[ 0.68637615 -0.7386913 -0.513895 0.87293863 -0.32614982 -1.0020137\n", + " 0.7798594 0.5945897 -0.05700803 -1.0218383 -0.144294 0.9858737\n", + " 0.33982527 -0.9821772 0.94524556 -0.38693175 0.85172147 -0.5121361 ] [ 0.63911194 -0.8075297 -0.6740009 0.76427484 -0.14366724 -1.0084257\n", + " 0.9317806 0.25870264 0.01909981 -1.0435557 -0.09680636 1.008331\n", + " 0.3066705 -0.9856341 0.7227781 -0.7072558 0.4838629 -0.9065276 ]\n", + "324.58446153766187 448.9328644285597\n", + "443.305059893921 352.68389964729084\n", + "378.6184367383896 369.389432527288\n", + "428.0366647294901 411.149494611942\n", + "378.958682516889 375.7145322241912\n", + "316.5447521996646 406.56681547772473\n", + "391.90051235819374 401.6722995774427\n", + "396.2088671051698 386.63109634857193\n", + "--------------------\n", + "Coordinate hypotheses and weights: [((350.1992483074778, 422.69904000881047), 6656), ((345.4704516491785, 426.75441952985426), 7107), ((348.35597040062885, 429.86029550564785), 7474), ((350.25719540946295, 421.5517480113508), 6501), ((346.9461762922956, 429.702904302672), 7424), ((348.4936986216981, 425.5558447624296), 7124), ((348.2289974631019, 427.44571449240226), 7533), ((353.8362114629093, 420.64495639502115), 6227), ((352.3014150072906, 423.00852629631095), 6545), ((348.3249302809231, 427.76149548191535), 7562), ((347.34749839416867, 427.8520028138533), 7502), ((347.3885905836952, 429.32065578951324), 7485), ((352.77979847386916, 421.8160782823978), 6413), ((346.4388551272584, 430.56226972100296), 7302), ((347.36313351820553, 424.39808052541616), 6932), ((353.8127803038824, 426.4993250046738), 6250), ((353.20730950189545, 424.7640780402657), 6453), ((360.8698489568442, 420.1925300164815), 5226), ((348.91649135600005, 429.6507100444274), 7485), ((346.94945230783196, 429.596460393803), 7432), ((347.06400727005337, 428.15001916162), 7489), ((352.9623198277364, 423.38939152248463), 6465), ((357.0524041924426, 418.7891223450368), 5788), ((349.231683521185, 418.28244615923984), 5884), ((348.15776898092463, 430.65194742574613), 7388), ((347.2349101241392, 427.64049273007345), 7468), ((350.25882186176716, 427.41278015416566), 7330), ((345.3618390117351, 426.46747472846897), 7054), ((348.21928299967533, 429.8590305736231), 7469), ((352.3278083066798, 424.3366241660095), 6618), ((362.905877992321, 411.2958423135703), 4518), ((346.14860568717677, 428.8766338269916), 7382), ((346.9152799735931, 426.9764007421109), 7351), ((349.59960616429225, 427.6843472137139), 7495), ((345.77745344121115, 426.84935778234967), 7161), ((330.3909372467104, 448.06368026971194), 5738), ((347.5471062511042, 428.1823958713806), 7526), ((338.7476036765961, 433.9034351143214), 6220), ((353.59413755014737, 422.77512205416815), 6348), ((349.78823027142255, 427.20945987008946), 7393), ((347.9391402567885, 424.57250721736057), 6974), ((316.6875934779827, 513.1606640307684), 1518), ((347.90787665585253, 431.16850084088753), 7327), ((344.4042474150225, 431.4306573952846), 6947), ((347.0247852885084, 427.3296206335037), 7415), ((347.91073935142526, 432.2308276275328), 7212), ((352.0990499782109, 422.1866593054095), 6531), ((324.58446153766187, 448.9328644285597), 5742)]\n", + "# Coordinate hypotheses and weights: 48\n" + ] + } + ], + "source": [ + "def ransacVal(y1, x1, v2): # dot product of unit vectors to find cos(theta difference)\n", + "\tv2 = v2 / np.linalg.norm(v2)\n", + "\t\n", + "\treturn y1 * v2[1] + x1 * v2[0]\n", + "\n", + "hypDict = {0: [], 1: [], 2: [], 3: [], 4: [], 5: [], 6: [], 7: [], 8: []}\n", + "\n", + "for n in range(50): #take two pixels, find intersection of unit vectors\n", + " #print(n)\n", + " p1 = population.pop(random.randrange(len(population)))\n", + " v1 = vecPred[p1[0]][p1[1]]\n", + " p2 = population.pop(random.randrange(len(population)))\n", + " v2 = vecPred[p2[0]][p2[1]]\n", + " print(p1, p2)\n", + " print(v1, v2)\n", + " for i in range(9): # find lines intersection, use as hypothesis\n", + " m1 = v1[i * 2 + 1] / v1[i * 2]\n", + " m2 = v2[i * 2 + 1] / v2[i * 2]\n", + " b1 = p1[0] - p1[1] * m1\n", + " b2 = p2[0] - p2[1] * m2\n", + " x = (b2 - b1) / (m1 - m2)\n", + " y = m1 * x + b1\n", + " if (y >= p1[0] != v1[i * 2 + 1] < 0 or x >= p1[1] != v1[i * 2] < 0 or y >= p2[0] != v2[i * 2 + 1] < 0 or x >= p2[1] != v2[i * 2] < 0) or not (m1 - m2): # check if line intersection takes place according to unit vector directions\n", + " continue\n", + " print(y, x)\n", + " weight = 0\n", + " for voter in population: # voting for fit of hypothesis\n", + " yDiff = y - voter[0]\n", + " xDiff = x - voter[1]\n", + "\n", + " mag = math.sqrt(yDiff ** 2 + xDiff ** 2)\n", + " vec = vecPred[voter[0]][voter[1]][i * 2: i * 2 + 2]\n", + "\n", + " if ransacVal(yDiff / mag, xDiff / mag, vec) > .99:\n", + " weight += 1\n", + " hypDict[i].append(((y, x), weight))\n", + "\n", + " population.append(p1)\n", + " population.append(p2)\n", + " print(\"--------------------\")\n", + "\n", + "print(\"Coordinate hypotheses and weights: \" + str(hypDict[0]))\n", + "print(\"# Coordinate hypotheses and weights: \" + str(len(hypDict[0])))" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Data Preparation\n", + "\n", + "Now we have a dictionary with a key / value pair for each keypoint. Each value contains a list of coordinates, each with an associated weight indicating the 'fit' of that keypoint to the rest of the popluation.\n", + "\n", + "Better performance can be achieved by pruning some of the outlying values from these hypotheses, one method is to find the mean, and standard deviation values for each keypoint, and prune those that lie outside a certain range." + ] + }, + { + "cell_type": "code", + "execution_count": 62, + "metadata": {}, + "outputs": [], + "source": [ + "def pruneHypsStdDev(hypDict, m = 2): # prune generated hypotheses using mean and stdDev\n", + "\tfor key, hyps in hypDict.items():\n", + "\t\tyVals, xVals = [x[0][0]for x in hyps], [x[0][1]for x in hyps]\n", + "\t\tyMean, xMean = statistics.mean(yVals), statistics.mean(xVals)\n", + "\t\tyDev, xDev = statistics.pstdev(yVals) * m, statistics.pstdev(xVals) * m\n", + "\t\thypDict[key] = [x for x in hyps if not determineOutlier(x[0], yMean, yDev, xMean, xDev)]\n", + " \n", + "def determineOutlier(input, yMean, yDev, xMean, xDev):\n", + "\treturn abs(input[0] - yMean) > yDev or abs(input[1] - xMean) > xDev\n", + " \n", + "pruneHypsStdDev(hypDict)" + ] + }, + { + "cell_type": "code", + "execution_count": 63, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "# Coordinate hypotheses and weights: 44\n" + ] + } + ], + "source": [ + "print(\"# Coordinate hypotheses and weights: \" + str(len(hypDict[0])))" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Now we need our keypoints to use to solve our PnP problem. This can be done my taking a weighted average of the hypotheses for each of the keypoints." + ] + }, + { + "cell_type": "code", + "execution_count": 64, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "{0: [348.92349714421067, 426.71295612304755], 1: [446.07859347203174, 349.22886174352965], 2: [376.381928760519, 369.9785432177094], 3: [426.90189165090896, 408.3436822580204], 4: [382.7737493833683, 376.2693643580854], 5: [442.62946617368334, 370.9827536454358], 6: [339.06795209705604, 399.0731164189295], 7: [389.4221778264622, 406.60440837199707], 8: [393.16503317529725, 388.92481391417846]}\n" + ] + } + ], + "source": [ + "def getMean(hypDict): # get weighted average of coordinates\n", + "\tmeanDict = {}\n", + "\tfor key, hyps in hypDict.items():\n", + "\t\txMean = 0\n", + "\t\tyMean = 0\n", + "\t\ttotalWeight = 0\n", + "\t\tfor hyp in hyps:\n", + "\t\t\tyMean += hyp[0][0] * hyp[1]\n", + "\t\t\txMean += hyp[0][1] * hyp[1]\n", + "\t\t\ttotalWeight += hyp[1]\n", + "\t\tyMean /= totalWeight\n", + "\t\txMean /= totalWeight\n", + "\t\tmeanDict[key] = [yMean, xMean]\n", + "\treturn meanDict\n", + "\n", + "meanDict = getMean(hypDict)\n", + "\n", + "print(meanDict)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## PnP Solver\n", + "\n", + "With our 9 keypoints, we are now ready to solve the PnP problem. First we load the corresponding 3d keypoints for the 2d keypoints we have found, then we can use cv2 to find the camera position relative to the object." + ] + }, + { + "cell_type": "code", + "execution_count": 65, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "array([[-0.013912, 0.063992, 0.010415],\n", + " [ 0.006557, -0.051654, -0.008197],\n", + " [ 0.026384, 0.010894, 0.008823],\n", + " [-0.030676, -0.006337, 0.004477],\n", + " [ 0.002858, -0.007523, -0.032334],\n", + " [ 0.003312, -0.024687, 0.031944],\n", + " [ 0.007547, 0.040566, -0.016576],\n", + " [-0.01014 , 0.024009, 0.01699 ]])" + ] + }, + "execution_count": 65, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "pts3d = np.loadtxt(os.path.join(os.getcwd() + f'/LINEMOD/{class_name}/', 'farthest.txt'))\n", + "pts3d" + ] + }, + { + "cell_type": "code", + "execution_count": 66, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Rotation Vector: [[-0.46291758]\n", + " [-0.83103027]\n", + " [-2.44160156]]\n", + "Translation Vector: [[0.0629583 ]\n", + " [0.11551684]\n", + " [0.53120925]]\n" + ] + } + ], + "source": [ + "def dictToArray(hypDict): # take dictionary keypoints and return list object\n", + "\tcoordArray = np.zeros((len(hypDict.keys()), 2))\n", + "\tfor key, hyps in hypDict.items():\n", + "\t\tcoordArray[key] = np.array([round(hyps[1]), round(hyps[0])]) # x, y format\n", + "\treturn coordArray\n", + "\n", + "# pts3d = np.loadtxt(os.path.join(os.getcwd() + f'/LINEMOD/{class_name}/', 'corners.txt'))\n", + "preds = dictToArray(meanDict)[:8] # ignoring centroid prediction ******* ### [1:] ### ********\n", + "\n", + "# matrix = np.array([[572.4114, 0., 325.2611], [0., 573.57043, 242.04899], [0., 0., 1.]]) # input camera matrix for the kinect camera\n", + "matrix = np.array([[543.25272224, 0., 320.25], [0., 724.33696299, 240.33333333], [0., 0., 1.]]) # camera matrix GUIMOD\n", + "\n", + "import cv2\n", + "\n", + "_, rVec, tVec = cv2.solvePnP(pts3d, preds, matrix, np.zeros(shape=[8, 1], dtype='float64'), flags = cv2.SOLVEPNP_ITERATIVE)\n", + "\n", + "print(\"Rotation Vector: \" + str(rVec)) # output rotation vector\n", + "print(\"Translation Vector: \" + str(tVec)) # output translation vector" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Results\n", + "\n", + "Now that we have our rotation and translation vectors, we can perform a sort of inverse process and impose a 3d object on our 2d image using the cv2 projectPoints function. The arguments for this function are the 3D coordinates we wish to project on to our 2D image, along with our newly found rotation and translation vectors, as well as the camera matrix and distortion coefficients.\n", + "\n", + "We can then check the accuracy of our prediction using the some keras metrics functions comparing the predicted keypoint pixel values to their true values, and even draw the true and predicted bounding boxes on the original image to get a visual indication of our accuracy." + ] + }, + { + "cell_type": "code", + "execution_count": 72, + "metadata": {}, + "outputs": [], + "source": [ + "pts3d2 = np.loadtxt(os.path.join(os.getcwd() + f'/LINEMOD/{class_name}/', 'corners.txt'))" + ] + }, + { + "cell_type": "code", + "execution_count": 92, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Keypoint pixel predictions: [[390.56203346 338.40180408]\n", + " [445.13724767 373.76187332]\n", + " [324.95614689 466.06423036]\n", + " [381.96303262 513.21975156]\n", + " [387.11617784 290.76990737]\n", + " [446.81734251 326.77122535]\n", + " [314.30692666 425.84587836]\n", + " [376.89214864 475.52128796]]\n", + "Keypoint ground truth: [[390.69925727 340.15530006]\n", + " [444.72414651 373.20924977]\n", + " [328.32509903 467.61479117]\n", + " [384.80626016 512.01923182]\n", + " [386.00447201 293.76869805]\n", + " [444.90527597 327.26218047]\n", + " [316.78972231 428.56966271]\n", + " [378.57629839 475.17138205]]\n" + ] + }, + { + "data": { + "image/png": "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\n", + "text/plain": [ + "<Figure size 432x288 with 1 Axes>" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Prediction Accuracy - MAE: 1.6, MSE 3.59\n" + ] + } + ], + "source": [ + "yPred = cv2.projectPoints(pts3d2, rVec, tVec, matrix, np.zeros(shape=[8, 1], dtype='float64'))[0]\n", + "yPred = np.squeeze(yPred)\n", + "\n", + "# yTrue = data.labelFloatsToPixels(labels, decPlace = 8)[:8] # converting our labels to pixel values #####[1::]\n", + "pose = np.load(f'/home/mahmoud/stvNet/LINEMOD/{class_name}/pose/{int(img_id[0])}.npy')\n", + "rVec_True = np.array(pose[0:3, 0:3], dtype='float64')\n", + "tVec_True = np.array(pose[0:3, 3], dtype='float64')\n", + "\n", + "yTrue = cv2.projectPoints(pts3d2, rVec_True, tVec_True, matrix, np.zeros(shape=[8, 1], dtype='float64'))[0]\n", + "yTrue = np.squeeze(yTrue)\n", + "\n", + "print(\"Keypoint pixel predictions: \" + str(yPred))\n", + "print(\"Keypoint ground truth: \" + str(yTrue))\n", + "\n", + "\n", + "def labelDrawPoints(drawList): # (b, f = back, front), (l, r = left, right), (u, d = up , down)\n", + "\tdrawDict = {}\n", + "\tdrawDict['bld'] = (int(round(drawList[0][0])), int(round(drawList[0][1])))\n", + "\tdrawDict['blu'] = (int(round(drawList[1][0])), int(round(drawList[1][1])))\n", + "\tdrawDict['fld'] = (int(round(drawList[2][0])), int(round(drawList[2][1])))\n", + "\tdrawDict['flu'] = (int(round(drawList[3][0])), int(round(drawList[3][1])))\n", + "\tdrawDict['brd'] = (int(round(drawList[4][0])), int(round(drawList[4][1])))\n", + "\tdrawDict['bru'] = (int(round(drawList[5][0])), int(round(drawList[5][1])))\n", + "\tdrawDict['frd'] = (int(round(drawList[6][0])), int(round(drawList[6][1])))\n", + "\tdrawDict['fru'] = (int(round(drawList[7][0])), int(round(drawList[7][1])))\n", + "\treturn drawDict\n", + "\n", + "def drawPose(img, drawPoints, colour = (255,0,0)): # draw bounding box\n", + "\t\n", + "\tcv2.line(img, drawPoints['bld'], drawPoints['blu'], colour, 2)\n", + "\tcv2.line(img, drawPoints['bld'], drawPoints['fld'], colour, 2)\n", + "\tcv2.line(img, drawPoints['bld'], drawPoints['brd'], colour, 2)\n", + "\tcv2.line(img, drawPoints['blu'], drawPoints['flu'], colour, 2)\n", + "\tcv2.line(img, drawPoints['blu'], drawPoints['bru'], colour, 2)\n", + "\tcv2.line(img, drawPoints['fld'], drawPoints['flu'], colour, 2)\n", + "\tcv2.line(img, drawPoints['fld'], drawPoints['frd'], colour, 2)\n", + "\tcv2.line(img, drawPoints['flu'], drawPoints['fru'], colour, 2)\n", + "\tcv2.line(img, drawPoints['fru'], drawPoints['bru'], colour, 2)\n", + "\tcv2.line(img, drawPoints['fru'], drawPoints['frd'], colour, 2)\n", + "\tcv2.line(img, drawPoints['frd'], drawPoints['brd'], colour, 2)\n", + "\tcv2.line(img, drawPoints['brd'], drawPoints['bru'], colour, 2)\n", + "\n", + "img = cv2.cvtColor(image, cv2.COLOR_RGB2BGR)\n", + "\n", + "drawPose(img, labelDrawPoints(yPred)) # drawing our predicted bounding box in blue\n", + "drawPose(img, labelDrawPoints(yTrue), (0,255,0)) # drawing the true bounding box in green\n", + "showImage(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))\n", + "\n", + "print(\"Prediction Accuracy - MAE: {0}, MSE {1}\".format(round(tf.reduce_mean(tf.keras.metrics.mean_absolute_error(yTrue, yPred)).numpy(), 2), round(tf.reduce_mean(tf.keras.metrics.mean_squared_error(yTrue, yPred)).numpy(), 2)))" + ] + }, + { + "cell_type": "code", + "execution_count": 83, + "metadata": {}, + "outputs": [ + { + "data": { + "image/png": "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\n", + "text/plain": [ + "<Figure size 432x288 with 1 Axes>" + ] + }, + "metadata": { + "needs_background": "light" + }, + "output_type": "display_data" + } + ], + "source": [ + "from test_vis import test_res\n", + "from plyfile import PlyData\n", + "\n", + "pose = np.load(f'/home/mahmoud/stvNet/LINEMOD/{class_name}/pose/{int(img_id[0])}.npy')\n", + "pc = f'/home/mahmoud/stvNet/LINEMOD/{class_name}/{class_name}.ply'\n", + "\n", + "plydata = PlyData.read(pc)\n", + "elm = plydata.elements\n", + "data = np.asarray(elm[0][:])\n", + "\n", + "res_pc = np.zeros((len(data), 3))\n", + "\n", + "for i in range(len(data)):\n", + " res_pc[i][0], res_pc[i][1], res_pc[i][2] = data[i][0], data[i][1], data[i][2]\n", + "\n", + "\n", + "test_res(image, res_pc, pose, rVec, tVec, matrix)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Evaluation methods\n", + "\n", + "- Evaluate PVNet method using two common metrics: \n", + " - 2D projection metric.\n", + " - Average 3D distance of model points (ADD) metric.\n", + " \n", + "- **2D Projection metric**: \n", + " - Computes the mean distance between the projections of 3D model points given the estimated and the ground truth pose. \n", + " - A pose is considered as correct if the distance is less than **5 pixels**.\n", + "\n", + "- **ADD metric**: \n", + " - Transform the model points by the estimated and the ground truth poses, respectively.\n", + " - Compute the mean distance between the two transformed point sets. \n", + " - When the distance is less than **10% of the model’s diameter**, it is claimed that the estimated pose is correct.\n" + ] + }, + { + "cell_type": "code", + "execution_count": 93, + "metadata": {}, + "outputs": [], + "source": [ + "def transform_pts_Rt(pts, R, t):\n", + " \"\"\"Applies a rigid transformation to 3D points.\n", + " :param pts: nx3 ndarray with 3D points.\n", + " :param R: 3x3 ndarray with a rotation matrix.\n", + " :param t: 3x1 ndarray with a translation vector.\n", + " :return: nx3 ndarray with transformed 3D points.\n", + " \"\"\"\n", + " assert (pts.shape[1] == 3)\n", + " pts_t = R.dot(pts.T) + t.reshape((3, 1))\n", + " return pts_t.T\n", + "\n", + "def project_pts(pts, K, R, t):\n", + " \"\"\"Projects 3D points.\n", + " :param pts: nx3 ndarray with the 3D points.\n", + " :param K: 3x3 ndarray with an intrinsic camera matrix.\n", + " :param R: 3x3 ndarray with a rotation matrix.\n", + " :param t: 3x1 ndarray with a translation vector.\n", + " :return: nx2 ndarray with 2D image coordinates of the projections.\n", + " \"\"\"\n", + " assert (pts.shape[1] == 3)\n", + " P = K.dot(np.hstack((R, t)))\n", + " pts_h = np.hstack((pts, np.ones((pts.shape[0], 1))))\n", + " pts_im = P.dot(pts_h.T)\n", + " pts_im /= pts_im[2, :]\n", + " return pts_im[:2, :].T\n", + "\n", + "\n", + "def add(R_est, t_est, R_gt, t_gt, pts):\n", + " \"\"\"Average Distance of Model Points for objects with no indistinguishable\n", + " views - by Hinterstoisser et al. (ACCV'12).\n", + " :param R_est: 3x3 ndarray with the estimated rotation matrix.\n", + " :param t_est: 3x1 ndarray with the estimated translation vector.\n", + " :param R_gt: 3x3 ndarray with the ground-truth rotation matrix.\n", + " :param t_gt: 3x1 ndarray with the ground-truth translation vector.\n", + " :param pts: nx3 ndarray with 3D model points.\n", + " :return: The calculated error.\n", + " \"\"\"\n", + " pts_est = transform_pts_Rt(pts, R_est, t_est)\n", + " pts_gt = transform_pts_Rt(pts, R_gt, t_gt)\n", + " e = np.linalg.norm(pts_est - pts_gt, axis=1).mean()\n", + " return e\n", + "\n", + "def re(R_est, R_gt):\n", + " \"\"\"Rotational Error.\n", + " :param R_est: 3x3 ndarray with the estimated rotation matrix.\n", + " :param R_gt: 3x3 ndarray with the ground-truth rotation matrix.\n", + " :return: The calculated error.\n", + " \"\"\"\n", + " error_cos = float(0.5 * (np.trace(R_est.dot(np.linalg.inv(R_gt))) - 1.0))\n", + "\n", + " # Avoid invalid values due to numerical errors.\n", + " error_cos = min(1.0, max(-1.0, error_cos))\n", + "\n", + " error = math.acos(error_cos)\n", + " error = 180.0 * error / np.pi # Convert [rad] to [deg].\n", + " return error\n", + "\n", + "\n", + "def te(t_est, t_gt):\n", + " \"\"\"Translational Error.\n", + " :param t_est: 3x1 ndarray with the estimated translation vector.\n", + " :param t_gt: 3x1 ndarray with the ground-truth translation vector.\n", + " :return: The calculated error.\n", + " \"\"\"\n", + " assert (t_est.size == t_gt.size == 3)\n", + " error = np.linalg.norm(t_gt - t_est)\n", + " return error\n", + "\n", + "\n", + "def proj(R_est, t_est, R_gt, t_gt, K, pts):\n", + " \"\"\"Average distance of projections of object model vertices [px]\n", + " - by Brachmann et al. (CVPR'16).\n", + " :param R_est: 3x3 ndarray with the estimated rotation matrix.\n", + " :param t_est: 3x1 ndarray with the estimated translation vector.\n", + " :param R_gt: 3x3 ndarray with the ground-truth rotation matrix.\n", + " :param t_gt: 3x1 ndarray with the ground-truth translation vector.\n", + " :param K: 3x3 ndarray with an intrinsic camera matrix.\n", + " :param pts: nx3 ndarray with 3D model points.\n", + " :return: The calculated error.\n", + " \"\"\"\n", + " proj_est = project_pts(pts, K, R_est, t_est)\n", + " proj_gt = project_pts(pts, K, R_gt, t_gt)\n", + " e = np.linalg.norm(proj_est - proj_gt, axis=1).mean()\n", + " return e\n", + "\n", + "\n", + "def cou_mask(mask_est, mask_gt):\n", + " \"\"\"Complement over Union of 2D binary masks.\n", + " :param mask_est: hxw ndarray with the estimated mask.\n", + " :param mask_gt: hxw ndarray with the ground-truth mask.\n", + " :return: The calculated error.\n", + " \"\"\"\n", + " mask_est_bool = mask_est.astype(bool)\n", + " mask_gt_bool = mask_gt.astype(bool)\n", + "\n", + " inter = np.logical_and(mask_gt_bool, mask_est_bool)\n", + " union = np.logical_or(mask_gt_bool, mask_est_bool)\n", + "\n", + " union_count = float(union.sum())\n", + " if union_count > 0:\n", + " e = 1.0 - inter.sum() / union_count\n", + " else:\n", + " e = 1.0\n", + " return e\n" + ] + }, + { + "cell_type": "code", + "execution_count": 95, + "metadata": {}, + "outputs": [], + "source": [ + "def read_3d_points_linemod(object_name):\n", + " filename = f'LINEMOD/{object_name}/{object_name}.ply'\n", + " with open(filename) as f:\n", + " in_vertex_list = False\n", + " vertices = []\n", + " in_mm = False\n", + " for line in f:\n", + " if in_vertex_list:\n", + " vertex = line.split()[:3]\n", + " vertex = np.array([float(vertex[0]),\n", + " float(vertex[1]),\n", + " float(vertex[2])], dtype=np.float32)\n", + " if in_mm:\n", + " vertex = vertex / np.float32(10) # mm -> cm\n", + " vertex = vertex / np.float32(100)\n", + " vertices.append(vertex)\n", + " if len(vertices) >= vertex_count:\n", + " break\n", + " elif line.startswith('element vertex'):\n", + " vertex_count = int(line.split()[-1])\n", + " elif line.startswith('end_header'):\n", + " in_vertex_list = True\n", + " elif line.startswith('element face'):\n", + " in_mm = True\n", + " return np.matrix(vertices)" + ] + }, + { + "cell_type": "code", + "execution_count": 94, + "metadata": {}, + "outputs": [], + "source": [ + "def read_diameter(object_name):\n", + " # this is the same for linemod and occlusion linemod\n", + " \n", + " filename = f'/home/mahmoud/stvNet/LINEMOD/{object_name}/diameter.txt'\n", + " with open(filename) as f:\n", + " diameter_in_cm = float(f.readline())\n", + " return diameter_in_cm * 0.01\n" + ] + }, + { + "cell_type": "code", + "execution_count": 96, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "0.12950802\n" + ] + } + ], + "source": [ + "# record = np.load(args.prediction_file, allow_pickle=True).item()\n", + "# pts3d = read_3d_points_linemod(class_name)\n", + "\n", + "diameter = read_diameter(class_name)\n", + "print(diameter)" + ] + }, + { + "cell_type": "code", + "execution_count": 99, + "metadata": {}, + "outputs": [], + "source": [ + "# Read Pose of the image \n", + "# Change Pose number based on the number of image\n", + "\n", + "pose = np.load(f'/home/mahmoud/stvNet/LINEMOD/{class_name}/pose/{int(img_id[0])}.npy')\n", + "rVec_gt = np.array(pose[0:3, 0:3], dtype='float64')\n", + "tVec_gt = np.array(pose[0:3, 3], dtype='float64')\n" + ] + }, + { + "cell_type": "code", + "execution_count": 100, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "(3, 1)" + ] + }, + "execution_count": 100, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "tVec_gt.shape\n", + "tVec_gt =tVec_gt.reshape(3,1)\n", + "tVec_gt.shape" + ] + }, + { + "cell_type": "code", + "execution_count": 123, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "(3, 3)" + ] + }, + "execution_count": 123, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "rVec_gt.shape" + ] + }, + { + "cell_type": "code", + "execution_count": 126, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "(3, 3)" + ] + }, + "execution_count": 126, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "# Comvert from vector to matrix \n", + "# src = 3*1\n", + "# cv2.Rodrigues2(rVec, rVec2, jacobian=0)\n", + "from scipy.spatial.transform import Rotation as R\n", + "\n", + "r = R.from_rotvec(rVec.reshape(3,))\n", + "rVec_est = np.array(r.as_matrix())\n", + "rVec_est.shape" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Evaluation result" + ] + }, + { + "cell_type": "code", + "execution_count": 132, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "0.011708668854770895" + ] + }, + "execution_count": 132, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "add(rVec_est, tVec, rVec_gt, tVec_gt, pts3d)" + ] + }, + { + "cell_type": "code", + "execution_count": 133, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "True" + ] + }, + "execution_count": 133, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "add(rVec_est, tVec, rVec_gt, tVec_gt, pts3d)< diameter*0.1" + ] + }, + { + "cell_type": "code", + "execution_count": 134, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "2.4777170427258417" + ] + }, + "execution_count": 134, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "proj(rVec_est, tVec, rVec_gt, tVec_gt, matrix, pts3d)" + ] + }, + { + "cell_type": "code", + "execution_count": 135, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "True" + ] + }, + "execution_count": 135, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "proj(rVec_est, tVec, rVec_gt, tVec_gt, matrix, pts3d)< 5" + ] + }, + { + "cell_type": "code", + "execution_count": 136, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "0.9751418182767693" + ] + }, + "execution_count": 136, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "cou_mask(classPred, mask)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "---" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Keypoint Selection\n", + "* Select **k** keypoint using FPS algorithm\n", + "1. Initialize keypoint set by adding the object center.\n", + "2. Repeatedly find a point on object surface farthest to current keypoint set.\n", + "3. Add this point to the set until the size of the set = k" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Multiple instances\n", + "- PVNet method can handle multiple instances.\n", + "1. For each object class, we generate the hypotheses of the object centers and their voting scores using our proposed voting scheme. \n", + "2. Then,find the modes among the hypotheses and mark these modes as centers of different instances. \n", + "3. Finally, the instance masks are obtained by assigning pixels to the nearest instance center they vote for." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Symmetric objects\n", + "* There are ambiguities of keypoint locations. \n", + "* To eliminate the ambiguities, rotate the symmetric object to a canonical pose during training, as suggested by [**Bb8:** A scalable, accurate, robust to partial occlusion method for predicting the 3d poses of]" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "## Conclusion\n", + "\n", + "This gives us our final output values for this process, and can be used to generated a number of accuracy metrics, as well as the photo rendered above which gives a visual representation of the accuracy of the process." + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 3", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.7.8" + } + }, + "nbformat": 4, + "nbformat_minor": 2 +} -- GitLab