test-ICVL.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### ICVL"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "root = '/home/wuyiming/git/Hand'\n",
    "caffe_root = root + '/caffe'\n",
    "import sys\n",
    "sys.path.insert(0, caffe_root + '/python')\n",
    "sys.path.insert(0, 'lib')\n",
    "sys.path.insert(0, 'lib/data_layer/')\n",
    "sys.path.insert(0, 'lib/data/')\n",
    "sys.path.insert(0, 'lib/util/')\n",
    "\n",
    "import caffe\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "import h5py\n",
    "from matplotlib.patches import Circle\n",
    "import scipy.io as scio\n",
    "import os\n",
    "import time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "fx = 241.42\n",
    "fy = 241.42\n",
    "ux = 160\n",
    "uy = 120\n",
    "def jointsImgTo3D(sample):\n",
    "    \"\"\"\n",
    "    Normalize sample to metric 3D\n",
    "    :param sample: joints in (x,y,z) with x,y in image coordinates and z in mm\n",
    "    :return: normalized joints in mm\n",
    "    \"\"\"\n",
    "    ret = np.zeros((sample.shape[0], 3), np.float32)\n",
    "    for i in range(sample.shape[0]):\n",
    "        ret[i] = jointImgTo3D(sample[i])\n",
    "    return ret\n",
    "\n",
    "def jointImgTo3D(sample):\n",
    "    \"\"\"\n",
    "    Normalize sample to metric 3D\n",
    "    :param sample: joints in (x,y,z) with x,y in image coordinates and z in mm\n",
    "    :return: normalized joints in mm\n",
    "    \"\"\"\n",
    "    ret = np.zeros((3,), np.float32)\n",
    "    # convert to metric using f, see Thomson et al.\n",
    "    ret[0] = (sample[0] - ux) * sample[2] / fx\n",
    "    ret[1] = (sample[1] - uy) * sample[2] / fy\n",
    "    ret[2] = sample[2]\n",
    "    return ret\n",
    "\n",
    "def loadPredFile(filepath):\n",
    "    import os\n",
    "    assert os.path.isfile(filepath), \"{} is not exists or is not a file!\".format(filepath)\n",
    "    with open(filepath, 'r') as f:\n",
    "        lines = f.readlines()\n",
    "        for index, line in enumerate(lines):\n",
    "            lines[index] = map(float, line.split())\n",
    "        joints = np.array(lines)\n",
    "        print joints.shape\n",
    "        n, d = joints.shape\n",
    "        return joints.reshape(n, d/3, 3)\n",
    "    \n",
    "def getNumFrameWithinMaxDist(gt, joints, dist):\n",
    "    return (np.nanmax(np.sqrt(np.square(gt - joints).sum(axis=2)), axis=1) <= dist).sum()\n",
    "\n",
    "def plotAccuracy(gt, joints, thresh = 80, label='ours'):\n",
    "    fig = plt.figure()\n",
    "    ax = fig.add_subplot(111)\n",
    "    ax.plot([getNumFrameWithinMaxDist(gt, joints, j)/ float(joints.shape[0]) * 100. for j in range(0, 80)], label=label)\n",
    "    ax.grid(True)\n",
    "    plt.xlabel('Distance threshold / mm')\n",
    "    plt.ylabel('Fraction of frames within threshold / %')\n",
    "    plt.xlim((0, 80))\n",
    "    plt.ylim((0.0, 100.0))\n",
    "\n",
    "def predict_joints(model_name, weights_num):\n",
    "    \"\"\"predict joints\"\"\"\n",
    "    model_def = 'models/ICVL/hand_' + model_name + '/hand_' + model_name + '.prototxt'\n",
    "    model_weights = 'weights/ICVL/hand_' + model_name + '/hand_' + model_name + '_iter_' + weights_num + '.caffemodel'\n",
    "\n",
    "    print model_def\n",
    "    print model_weights\n",
    "    \n",
    "    net = caffe.Net(model_def,\n",
    "                   model_weights,\n",
    "                   caffe.TEST)\n",
    "    \n",
    "    file_name = 'result/OURS/ICVL/hand_' + model_name + '_' + weights_num + '.txt'\n",
    "    \n",
    "    import os\n",
    "    if os.path.isfile(file_name):\n",
    "        return file_name\n",
    "    \n",
    "    print file_name\n",
    "    \n",
    "    t_start = time.time()\n",
    "\n",
    "    predicted_joints = np.array([None]* 702)\n",
    "    \n",
    "    if model_name == 'baseline':\n",
    "        frame_size, joint_size, = net.blobs['joint_pred'].data.shape\n",
    "        seq_size = 1\n",
    "    else:\n",
    "        frame_size, seq_size, joint_size = net.blobs['pred_joint'].data.shape\n",
    "    dim = 3\n",
    "    \n",
    "    for i in xrange(np.int(np.ceil(702./ (frame_size * seq_size)))):\n",
    "        net.forward()\n",
    "        print 'test iter = ', i\n",
    "        for j, ind in enumerate(net.blobs['inds'].data):\n",
    "            row = j / seq_size\n",
    "            col = j % seq_size\n",
    "            if predicted_joints[int(ind)] == None:\n",
    "                predicted_joints[int(ind)] = (net.blobs['pred_joint'].data[row][col].reshape(16, 3) \\\n",
    "                                              * net.blobs['config'].data[j][0] / 2 \\\n",
    "                                              + net.blobs['com'].data[j].reshape(1, 3)).copy()\n",
    "    t_end = time.time()\n",
    "    print 'time elapse {}'.format((t_end - t_start) / 702)\n",
    "    \n",
    "    with open(file_name, 'w') as f:\n",
    "        for i in xrange(predicted_joints.shape[0]):\n",
    "            for item in predicted_joints[i].reshape(16*3):\n",
    "                f.write(\"%s \"% item)\n",
    "            f.write(\"\\n\")\n",
    "     \n",
    "    return file_name\n",
    "\n",
    "def vis_square(data):\n",
    "    \"\"\"Take an array of shape (n, height, width) or (n, height, width, 3)\n",
    "       and visualize each (height, width) thing in a grid of size approx. sqrt(n) by sqrt(n)\"\"\"\n",
    "    \n",
    "    # normalize data for display\n",
    "    data = (data - data.min()) / (data.max() - data.min())\n",
    "    \n",
    "    # force the number of filters to be square\n",
    "    n = int(np.ceil(np.sqrt(data.shape[0])))\n",
    "    padding = (((0, n ** 2 - data.shape[0]),\n",
    "               (0, 1), (0, 1))                 # add some space between filters\n",
    "               + ((0, 0),) * (data.ndim - 3))  # don't pad the last dimension (if there is one)\n",
    "    data = np.pad(data, padding, mode='constant', constant_values=1)  # pad with ones (white)\n",
    "\n",
    "    # tile the filters into an image\n",
    "    data = data.reshape((n, n) + data.shape[1:]).transpose((0, 2, 1, 3) + tuple(range(4, data.ndim + 1)))\n",
    "    data = data.reshape((n * data.shape[1], n * data.shape[3]) + data.shape[4:])\n",
    "    \n",
    "    plt.imshow(data); plt.axis('off')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "size of dataset is test1: 702\n",
      "dataset: ICVL phase: test_1\n",
      "Outputs:  ['depth', 'joint', 'clip_markers', 'com', 'config', 'inds']\n"
     ]
    }
   ],
   "source": [
    "caffe.set_device(0)\n",
    "caffe.set_mode_gpu()\n",
    "net = caffe.Net('./models/ICVL/hand_lstm_small_frame_size/hand_lstm_small_frame_size.prototxt', \n",
    "                './weights/ICVL/hand_lstm_small_frame_size/hand_lstm_small_frame_size_iter_200000.caffemodel',\n",
    "                caffe.TEST)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Read ground truth"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "gt_file = './dataset/ICVL/test_seq_1.txt'\n",
    "gt = []\n",
    "f = open(gt_file, 'r')\n",
    "lines = f.readlines()\n",
    "for line in lines:\n",
    "    gt.append(map(float, line.split(' ')[1:-1]))\n",
    "gt = np.array(gt)\n",
    "gt3D = []\n",
    "for i in xrange(gt.shape[0]):\n",
    "    gt3D.append(jointsImgTo3D(gt[i].reshape(16, 3)))\n",
    "gt3D = np.array(gt3D)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[  30.87788582,   38.85021591,  368.85400391],\n",
       "       [   6.77187777,   51.50400925,  355.86999512],\n",
       "       [ -20.13546371,   34.17987442,  354.85101318],\n",
       "       [ -38.90395737,   18.12351799,  349.55499268],\n",
       "       [   5.42264652,  -11.51812172,  357.78500366],\n",
       "       [  -2.93231583,  -37.45681   ,  347.35998535],\n",
       "       [  -8.47680283,  -55.2942009 ,  343.59799194],\n",
       "       [  27.84412956,  -15.1800518 ,  363.06399536],\n",
       "       [  30.44153023,  -48.11777878,  355.29098511],\n",
       "       [  32.33983612,  -69.76394653,  353.11999512],\n",
       "       [  45.66884232,   -4.60396385,  365.38101196],\n",
       "       [  54.80501938,  -33.34441376,  359.79299927],\n",
       "       [  61.16431046,  -52.92427444,  357.46798706],\n",
       "       [  63.34932327,   10.08257389,  369.03201294],\n",
       "       [  79.11164856,   -7.20125055,  367.94198608],\n",
       "       [  90.86534119,  -20.93623352,  365.125     ]], dtype=float32)"
      ]
     },
     "execution_count": 140,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gt3D[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[  30.4537 ,   31.5554 ,  367.849  ],\n",
       "       [   1.78531,   43.6418 ,  353.589  ],\n",
       "       [ -28.5635 ,   26.7316 ,  353.695  ],\n",
       "       [ -48.611  ,   11.9366 ,  350.03   ],\n",
       "       [   2.3439 ,  -21.4219 ,  358.6    ],\n",
       "       [  -5.56242,  -44.9381 ,  348.304  ],\n",
       "       [ -10.6416 ,  -57.7374 ,  343.255  ],\n",
       "       [  27.3949 ,  -23.0857 ,  366.587  ],\n",
       "       [  31.1486 ,  -54.2314 ,  358.09   ],\n",
       "       [  34.0692 ,  -71.8086 ,  351.319  ],\n",
       "       [  46.8118 ,  -12.0764 ,  367.893  ],\n",
       "       [  54.7152 ,  -41.405  ,  361.893  ],\n",
       "       [  60.4714 ,  -58.1645 ,  355.642  ],\n",
       "       [  68.2592 ,    5.5283 ,  370.811  ],\n",
       "       [  83.914  ,  -10.5339 ,  368.719  ],\n",
       "       [  95.3923 ,  -19.0521 ,  360.095  ]])"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_joints[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### plot by ourselves"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from util.handpose_evaluation import ICVLHandposeEvaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "models/ICVL/hand_lstm_small_frame_size/hand_lstm_small_frame_size.prototxt\n",
      "weights/ICVL/hand_lstm_small_frame_size/hand_lstm_small_frame_size_iter_200000.caffemodel\n",
      "size of dataset is test1: 702\n",
      "dataset: ICVL phase: test_1\n",
      "Outputs:  ['depth', 'joint', 'clip_markers', 'com', 'config', 'inds']\n",
      "(702, 48)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEKCAYAAADjDHn2AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJi4XU0oAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f92ed8ebad0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "model_num = 'lstm_small_frame_size'\n",
    "weights_num = '200000'\n",
    "pred_ours_file = predict_joints(model_num, weights_num)\n",
    "pred_joints = loadPredFile(pred_ours_file)\n",
    "plotAccuracy(gt3D, pred_joints, label='ours')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}