local-evaluation-single-step.ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"local-evaluation-single-step.ipynb","provenance":[],"collapsed_sections":[]},"kernelspec":{"name":"python3","display_name":"Python 3"}},"cells":[{"cell_type":"code","metadata":{"id":"-U-R36obg151","executionInfo":{"status":"ok","timestamp":1602907585488,"user_tz":360,"elapsed":2697,"user":{"displayName":"victor sergio peñaloza","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhvDieTYec1JQengOvALsshqIS3yX453P4KflPZ5A=s64","userId":"17418028661329428856"}}},"source":["import tensorflow as tf"],"execution_count":1,"outputs":[]},{"cell_type":"code","metadata":{"id":"J_LAfh3HhDRy","executionInfo":{"status":"ok","timestamp":1602907585494,"user_tz":360,"elapsed":2063,"user":{"displayName":"victor sergio peñaloza","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhvDieTYec1JQengOvALsshqIS3yX453P4KflPZ5A=s64","userId":"17418028661329428856"}},"outputId":"acb1c37e-176e-48e9-aba0-6bbaf3948382","colab":{"base_uri":"https://localhost:8080/","height":34}},"source":["from google.colab import drive\n","drive.mount('/content/drive')"],"execution_count":2,"outputs":[{"output_type":"stream","text":["Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"0wvyVv6LhG50","executionInfo":{"status":"ok","timestamp":1602907585495,"user_tz":360,"elapsed":948,"user":{"displayName":"victor sergio peñaloza","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhvDieTYec1JQengOvALsshqIS3yX453P4KflPZ5A=s64","userId":"17418028661329428856"}}},"source":["# path variables\n","\n","base_path = \"/content/drive/My Drive/RESEARCH/2020/Pedestrian_prediction_simposio/\"\n"],"execution_count":3,"outputs":[]},{"cell_type":"code","metadata":{"id":"Lb9VYBm6hLkc","executionInfo":{"status":"ok","timestamp":1602907585739,"user_tz":360,"elapsed":283,"user":{"displayName":"victor sergio peñaloza","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhvDieTYec1JQengOvALsshqIS3yX453P4KflPZ5A=s64","userId":"17418028661329428856"}}},"source":["\n","ground_truth = \"/content/drive/My Drive/RESEARCH/2020/Pedestrian_prediction_simposio/data/pruebas-imagenes/students001.txt\"\n","predicted_0 = \"/content/drive/My Drive/RESEARCH/2020/Pedestrian_prediction_simposio/data/pruebas-imagenes/0/students001.txt.out\"\n","predicted_128 = \"/content/drive/My Drive/RESEARCH/2020/Pedestrian_prediction_simposio/data/pruebas-imagenes/128/students001.txt.out\"\n","predicted_256 = \"/content/drive/My Drive/RESEARCH/2020/Pedestrian_prediction_simposio/data/pruebas-imagenes/256/students001.txt.out\"\n","predicted_512 = \"/content/drive/My Drive/RESEARCH/2020/Pedestrian_prediction_simposio/data/pruebas-imagenes/512/students001.txt.out\"\n"],"execution_count":4,"outputs":[]},{"cell_type":"code","metadata":{"id":"5X2gNioa0wcE","executionInfo":{"status":"ok","timestamp":1602907588503,"user_tz":360,"elapsed":1346,"user":{"displayName":"victor sergio peñaloza","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhvDieTYec1JQengOvALsshqIS3yX453P4KflPZ5A=s64","userId":"17418028661329428856"}}},"source":["import pandas as pd\n","\n","# Create an empty Dataframe without any column names, indices or data\n","dataframe_gt = pd.read_csv(ground_truth, sep=' ',header=None, names = ['time','id', 'x', 'y'])\n","dataframe_pred_0 = pd.read_csv(predicted_0, sep=' ',header=None, names = ['time','id', 'x', 'y'])\n","dataframe_pred_128 = pd.read_csv(predicted_128, sep=' ',header=None, names = ['time','id', 'x', 'y'])\n","dataframe_pred_256 = pd.read_csv(predicted_256, sep=' ',header=None, names = ['time','id', 'x', 'y'])\n","dataframe_pred_512 = pd.read_csv(predicted_512, sep=' ',header=None, names = ['time','id', 'x', 'y'])\n"],"execution_count":5,"outputs":[]},{"cell_type":"code","metadata":{"id":"LlzYwCy_J--Q","executionInfo":{"status":"ok","timestamp":1602907589711,"user_tz":360,"elapsed":973,"user":{"displayName":"victor sergio peñaloza","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhvDieTYec1JQengOvALsshqIS3yX453P4KflPZ5A=s64","userId":"17418028661329428856"}}},"source":["import math\n","\n","def l2_distance(gx,px,gy,py):\n","     return math.sqrt (((px - gx) ** 2) + ((py - gy) ** 2))"],"execution_count":6,"outputs":[]},{"cell_type":"code","metadata":{"id":"q3pZXeKRFrZA","executionInfo":{"status":"ok","timestamp":1602900932801,"user_tz":360,"elapsed":11635,"user":{"displayName":"victor sergio peñaloza","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhvDieTYec1JQengOvALsshqIS3yX453P4KflPZ5A=s64","userId":"17418028661329428856"}},"outputId":"16b7c9e8-6167-42c1-dfe8-9948aed7fdd3","colab":{"base_uri":"https://localhost:8080/","height":85}},"source":["total_mean = 0\n","\n","for row in range(0, len(dataframe_gt.index)):\n","    total_mean += l2_distance(dataframe_gt.iloc[row,:].x,dataframe_pred.iloc[row,:].x,dataframe_gt.iloc[row,:].y,dataframe_pred.iloc[row,:].y)\n","\n","total_mean = total_mean / len(dataframe_gt.index)\n","\n","total_final = 0\n","\n","for row in range(19,len(dataframe_gt.index),20):\n","    total_final += l2_distance(dataframe_gt.iloc[row,:].x,dataframe_pred.iloc[row,:].x,dataframe_gt.iloc[row,:].y,dataframe_pred.iloc[row,:].y)\n","\n","total_final = total_final / (len(dataframe_gt.index)/20)\n","\n","print(\"total mean distance: {:f} \\n\".format(total_mean))\n","print(\"total final displacement: {:f} \\n\".format(total_final))\n"],"execution_count":170,"outputs":[{"output_type":"stream","text":["total mean distance: 0.482743 \n","\n","total final displacement: 1.398008 \n","\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"JrGicuFFRE9E","executionInfo":{"status":"ok","timestamp":1602907593750,"user_tz":360,"elapsed":1274,"user":{"displayName":"victor sergio peñaloza","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhvDieTYec1JQengOvALsshqIS3yX453P4KflPZ5A=s64","userId":"17418028661329428856"}}},"source":["_gt = dataframe_gt.groupby(['id'])\n","_pred_0 = dataframe_pred_0.groupby(['id'])\n","_pred_128 = dataframe_pred_128.groupby(['id'])\n","_pred_256 = dataframe_pred_256.groupby(['id'])\n","_pred_512 = dataframe_pred_512.groupby(['id'])\n"],"execution_count":7,"outputs":[]},{"cell_type":"code","metadata":{"id":"zs51VyZfRxnI","executionInfo":{"status":"ok","timestamp":1602907610617,"user_tz":360,"elapsed":616,"user":{"displayName":"victor sergio peñaloza","photoUrl":"https://lh3.googleusercontent.com/a-/AOh14GhvDieTYec1JQengOvALsshqIS3yX453P4KflPZ5A=s64","userId":"17418028661329428856"}},"outputId":"c56d06be-f008-48cd-e520-b7e2291075a1","colab":{"base_uri":"https://localhost:8080/","height":401}},"source":["import matplotlib.pyplot as plt\n","\n","\n","pedestrian = 52  #184,195\n","\n","gx = _gt.get_group(pedestrian).x\n","gy = _gt.get_group(pedestrian).y\n","\n","px_0 = _pred_0.get_group(pedestrian).x\n","py_0 = _pred_0.get_group(pedestrian).y\n","\n","px_128 = _pred_128.get_group(pedestrian).x\n","py_128 = _pred_128.get_group(pedestrian).y\n","\n","px_256 = _pred_256.get_group(pedestrian).x\n","py_256 = _pred_256.get_group(pedestrian).y\n","\n","px_512 = _pred_512.get_group(pedestrian).x\n","py_512 = _pred_512.get_group(pedestrian).y\n","\n","\n","\n","gx = gx.to_numpy()\n","gy = gy.to_numpy()\n","\n","px_0 = px_0.to_numpy()\n","py_0 = py_0.to_numpy()\n","\n","px_128 = px_128.to_numpy()\n","py_128 = py_128.to_numpy()\n","\n","px_256 = px_256.to_numpy()\n","py_256 = py_256.to_numpy()\n","\n","px_512 = px_512.to_numpy()\n","py_512 = py_512.to_numpy()\n","\n","\n","\n","def print_metrics(gx, gy, px,py ):\n","    single_l2_distance = 0\n","\n","    for i in range (0,20):\n","        single_l2_distance += l2_distance(gx[i],px[i],gy[i],py[i])\n","\n","    single_l2_distance = single_l2_distance / 20\n","\n","    single_final_displacement = l2_distance(gx[19],px[19],gy[19],py[19])\n","\n","    print (\"Point mean distance error: \",single_l2_distance)\n","    print(\"Point final displacement error: \", single_final_displacement)\n","\n","\n","print_metrics(gx,gy,px_0,py_0)\n","print_metrics(gx,gy,px_128,py_128)\n","print_metrics(gx,gy,px_256,py_256)\n","print_metrics(gx,gy,px_512,py_512)\n","\n","\n","# Get current size\n","#fig_size = plt.rcParams[\"figure.figsize\"]\n"," \n","# Prints: [8.0, 6.0]\n","#print (\"Current size:\", fig_size)\n"," \n","# Set figure width to 12 and height to 9\n","#fig_size[0] = 12\n","#fig_size[1] = 9\n","#plt.rcParams[\"figure.figsize\"] = fig_size\n","    \n","    \n","#img = plt.imread(base_path+\"data/background/bookstore_video0.jpg\")\n","fig, ax = plt.subplots()\n","    \n","#x = range(300)\n","#ax.imshow(img, extent=[0, 400, 0, 300])\n","#ax.imshow(img)    \n","#ax.style.use('seaborn-whitegrid')\n","ax.plot(gx,gy,'-o',color='red',linestyle='dashed')\n","ax.plot(px_0,py_0,'-o',color='green',)\n","ax.plot(px_128,py_128,'-o',color='blue',)\n","ax.plot(px_256,py_256,'-o',color='black',)\n","ax.plot(px_512,py_512,'-o',color='purple',)\n","\n","plt.show()\n"],"execution_count":8,"outputs":[{"output_type":"stream","text":["Point mean distance error:  0.3067383797251816\n","Point final displacement error:  1.0548677642245026\n","Point mean distance error:  0.13863877387269236\n","Point final displacement error:  0.01004987562112071\n","Point mean distance error:  0.12086243374651759\n","Point final displacement error:  0.516400038729666\n","Point mean distance error:  0.12647359936062702\n","Point final displacement error:  0.5375295340723143\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[],"needs_background":"light"}}]},{"cell_type":"code","metadata":{"id":"Wacq5N3qdXKr"},"source":[""],"execution_count":null,"outputs":[]}]}