Skip to content

Instantly share code, notes, and snippets.

@hgaiser

hgaiser/TwoClass.ipynb

Created September 1, 2017 21:40
Show Gist options
  • Star 0 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save hgaiser/459dbe5f9020c8450dc87f417b6c877b to your computer and use it in GitHub Desktop.
Save hgaiser/459dbe5f9020c8450dc87f417b6c877b to your computer and use it in GitHub Desktop.
Download ZIP
Display the source blob
Display the rendered blob
Raw
TwoClass.ipynb
{
"cells": [
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"The autoreload extension is already loaded. To reload it, use:\n",
" %reload_ext autoreload\n"
]
}
],
"source": [
"%matplotlib inline\n",
"%load_ext autoreload\n",
"%autoreload 2\n",
"\n",
"import pdb\n",
"\n",
"import keras\n",
"import keras.preprocessing.image\n",
"from keras.applications.imagenet_utils import get_file\n",
"\n",
"import keras_resnet.models\n",
"import keras_retinanet.layers\n",
"import keras_retinanet.models\n",
"\n",
"import matplotlib.pyplot as plt\n",
"import cv2\n",
"import numpy as np\n",
"import sys\n",
"import math\n",
"\n",
"import tensorflow as tf\n",
"\n",
"def get_session():\n",
" config = tf.ConfigProto()\n",
" config.gpu_options.allow_growth = True\n",
" return tf.Session(config=config)\n",
"\n",
"keras.backend.tensorflow_backend.set_session(get_session())\n",
"\n",
"WEIGHTS_PATH_NO_TOP = 'https://github.com/fchollet/deep-learning-models/releases/download/v0.2/resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5'"
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/usr/lib/python3.6/site-packages/ipykernel/__main__.py:15: UserWarning: Output \"classification_softmax\" missing from loss dictionary. We assume this was done on purpose, and we will not be expecting any data to be passed to \"classification_softmax\" during training.\n",
"/usr/lib/python3.6/site-packages/ipykernel/__main__.py:15: UserWarning: Output \"concatenate_23\" missing from loss dictionary. We assume this was done on purpose, and we will not be expecting any data to be passed to \"concatenate_23\" during training.\n",
"/usr/lib/python3.6/site-packages/ipykernel/__main__.py:15: UserWarning: Output \"focal_loss_2\" missing from loss dictionary. We assume this was done on purpose, and we will not be expecting any data to be passed to \"focal_loss_2\" during training.\n",
"/usr/lib/python3.6/site-packages/ipykernel/__main__.py:15: UserWarning: Output \"concatenate_22\" missing from loss dictionary. We assume this was done on purpose, and we will not be expecting any data to be passed to \"concatenate_22\" during training.\n"
]
}
],
"source": [
"def create_model(num_classes=3, *args, **kwargs):\n",
" image = keras.layers.Input((512, 512, 3))\n",
" im_info = keras.layers.Input((3,))\n",
" gt_boxes = keras.layers.Input((None, 5))\n",
" \n",
" return keras_retinanet.models.ResNet50RetinaNet([image, im_info, gt_boxes], num_classes=num_classes)\n",
"\n",
"model = create_model()\n",
"\n",
"# load imagenet weights\n",
"weights_path = get_file('resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5', WEIGHTS_PATH_NO_TOP, cache_subdir='models', md5_hash='a268eb855778b3df3c7506639542a6af')\n",
"model.load_weights(weights_path, by_name=True)\n",
"\n",
"# compile model\n",
"model.compile(loss=None, optimizer=keras.optimizers.Adam(lr=1e-4))"
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAQYAAAD8CAYAAACVSwr3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x7f63e19c0d30>"
]
},
"metadata": {},
"output_type": "display_data"
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"(1, 512, 512, 3) (1, 3) (1, 2, 5)\n"
]
}
],
"source": [
"# generate really simple image with one object\n",
"image = np.zeros((512, 512, 3), dtype=keras.backend.floatx())\n",
"image[192:400, 192:400, :] = 1.0\n",
"image[64:128, 64:128, :] = -1.0\n",
"plt.imshow(image, cmap='gray')\n",
"plt.show()\n",
"\n",
"# create input batch blobs\n",
"im_info_batch = np.array([[512, 512, 1.0]])\n",
"gt_boxes_batch = np.array([[[192, 192, 400, 400, 2], [64, 64, 128, 128, 1]]])\n",
"\n",
"image_batch = np.expand_dims(image, axis=0)\n",
"\n",
"inputs = [image_batch, im_info_batch, gt_boxes_batch]\n",
"\n",
"def simple_data_generator():\n",
" while True:\n",
" yield inputs, None\n",
"\n",
"print(image_batch.shape, im_info_batch.shape, gt_boxes_batch.shape)"
]
},
{
"cell_type": "code",
"execution_count": 13,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"Epoch 1/10\n",
"100/100 [==============================] - 12s - loss: 4.7327e-07 \n",
"Epoch 2/10\n",
"100/100 [==============================] - 12s - loss: 3.9538e-07 \n",
"Epoch 3/10\n",
"100/100 [==============================] - 12s - loss: 3.3207e-07 \n",
"Epoch 4/10\n",
"100/100 [==============================] - 12s - loss: 2.8016e-07 \n",
"Epoch 5/10\n",
"100/100 [==============================] - 12s - loss: 2.3733e-07 \n",
"Epoch 6/10\n",
"100/100 [==============================] - 12s - loss: 2.0144e-07 \n",
"Epoch 7/10\n",
"100/100 [==============================] - 12s - loss: 1.7116e-07 \n",
"Epoch 8/10\n",
"100/100 [==============================] - 12s - loss: 1.4550e-07 \n",
"Epoch 9/10\n",
"100/100 [==============================] - 12s - loss: 1.2357e-07 \n",
"Epoch 10/10\n",
"100/100 [==============================] - 12s - loss: 1.0511e-07 \n"
]
},
{
"data": {
"text/plain": [
"<keras.callbacks.History at 0x7f645abee128>"
]
},
"execution_count": 13,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"# train for some iterations\n",
"model.fit_generator(generator=simple_data_generator(),steps_per_epoch=100,epochs=10,verbose=1)"
]
},
{
"cell_type": "code",
"execution_count": 14,
"metadata": {
"scrolled": false
},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"(1, 49104) (1, 49104, 3) (1, 49104, 4)\n",
"1.0 [ 2.07611942e-03 9.97923851e-01 8.24712035e-11]\n",
"1.0 [ 1.56506745e-03 9.98434961e-01 4.64021355e-09]\n",
"1.0 [ 2.26700786e-06 9.99997735e-01 1.67636641e-10]\n",
"1.0 [ 7.85623677e-04 9.99214411e-01 2.13077578e-09]\n",
"1.0 [ 2.11134943e-06 9.99997854e-01 4.09125755e-09]\n",
"1.0 [ 1.47294742e-03 9.98527050e-01 1.51459467e-08]\n",
"1.0 [ 9.21399312e-09 1.00000000e+00 1.45735035e-10]\n",
"1.0 [ 2.23192242e-09 1.00000000e+00 2.62091393e-09]\n",
"1.0 [ 1.32897890e-06 9.99998689e-01 4.00265110e-09]\n",
"1.0 [ 3.70260682e-06 9.99996305e-01 1.93765022e-08]\n",
"1.0 [ 2.19099573e-03 9.97808993e-01 3.85744636e-09]\n",
"1.0 [ 2.48410215e-04 9.99750555e-01 1.08024267e-06]\n",
"1.0 [ 4.45585293e-08 1.00000000e+00 3.21081717e-10]\n",
"1.0 [ 4.23887229e-07 9.99999166e-01 3.87737089e-07]\n",
"1.0 [ 1.50253763e-03 9.98493433e-01 4.05902438e-06]\n",
"1.0 [ 2.06248346e-03 9.97937560e-01 8.75598172e-09]\n",
"1.0 [ 1.04739307e-03 9.98952627e-01 2.19782628e-08]\n",
"1.0 [ 1.79870876e-05 9.99981999e-01 6.34386640e-13]\n",
"1.0 [ 3.37892375e-03 9.96621013e-01 2.71500265e-16]\n",
"1.0 [ 0.00169148 0.99830854 0. ]\n",
"1.0 [ 3.59520433e-03 9.96404767e-01 9.08144651e-22]\n",
"1.0 [ 2.18646252e-03 9.97813463e-01 5.99607495e-13]\n",
"1.0 [ 6.50533475e-05 9.99934912e-01 7.29690960e-12]\n",
"1.0 [ 1.28020400e-10 1.00000000e+00 0.00000000e+00]\n",
"1.0 [ 1.63928782e-09 1.00000000e+00 9.92936821e-17]\n",
"1.0 [ 1.15664232e-11 1.00000000e+00 3.42808979e-13]\n",
"1.0 [ 1.26417644e-11 1.00000000e+00 1.15590670e-13]\n",
"1.0 [ 2.32336230e-08 1.00000000e+00 0.00000000e+00]\n",
"1.0 [ 1.42165864e-05 9.99985814e-01 5.40237560e-12]\n",
"1.0 [ 0.00166771 0.99833232 0. ]\n",
"1.0 [ 3.04485881e-03 9.96955156e-01 5.53238255e-09]\n",
"1.0 [ 4.57107380e-04 9.99542952e-01 5.89325073e-12]\n",
"1.0 [ 2.95174832e-04 9.99704659e-01 1.71102045e-07]\n",
"1.0 [ 3.69009096e-03 9.96309936e-01 3.42398473e-13]\n",
"1.0 [ 1.53104274e-03 9.98468697e-01 2.21423718e-07]\n",
"1.0 [ 0.00146389 0.99853611 0. ]\n",
"1.0 [ 3.26309423e-03 9.96736825e-01 1.41052325e-07]\n",
"2.0 [ 1.42700563e-03 1.43542775e-10 9.98572946e-01]\n",
"2.0 [ 2.48760771e-04 3.27116148e-08 9.99751151e-01]\n",
"2.0 [ 4.18200689e-05 5.77188124e-12 9.99958158e-01]\n",
"2.0 [ 2.74131919e-04 5.03822619e-08 9.99725759e-01]\n",
"2.0 [ 1.12352578e-03 1.45903115e-07 9.98876274e-01]\n",
"2.0 [ 2.32492550e-03 4.49036224e-06 9.97670591e-01]\n",
"2.0 [ 2.57360842e-03 7.43304203e-12 9.97426331e-01]\n",
"2.0 [ 1.03544546e-04 4.95569509e-07 9.99895930e-01]\n",
"2.0 [ 1.41000934e-03 1.40816803e-06 9.98588502e-01]\n",
"2.0 [ 1.74566798e-04 1.26644972e-09 9.99825418e-01]\n",
"2.0 [ 5.69057993e-05 2.16667830e-12 9.99943137e-01]\n",
"2.0 [ 2.82478373e-04 4.96227448e-10 9.99717534e-01]\n",
"2.0 [ 1.45129266e-03 3.68589924e-12 9.98548687e-01]\n",
"2.0 [ 5.65189282e-07 2.36307415e-12 9.99999404e-01]\n",
"2.0 [ 3.13012949e-09 2.22377228e-09 1.00000000e+00]\n",
"2.0 [ 7.11576909e-10 1.17960561e-15 1.00000000e+00]\n",
"2.0 [ 1.43832363e-10 1.49871673e-12 1.00000000e+00]\n",
"2.0 [ 4.92735630e-08 8.37724225e-15 1.00000000e+00]\n",
"2.0 [ 1.67512528e-08 1.98524808e-10 1.00000000e+00]\n",
"2.0 [ 4.14237803e-07 1.54323885e-07 9.99999404e-01]\n",
"2.0 [ 6.17919284e-07 2.14155277e-15 9.99999404e-01]\n",
"2.0 [ 1.67996872e-09 2.96160006e-11 1.00000000e+00]\n",
"2.0 [ 1.58451451e-03 7.63657461e-14 9.98415470e-01]\n",
"2.0 [ 4.17562404e-07 3.08762438e-09 9.99999523e-01]\n",
"2.0 [ 2.69039022e-03 1.96636705e-07 9.97309446e-01]\n",
"2.0 [ 7.76574307e-04 1.72582656e-06 9.99221683e-01]\n",
"2.0 [ 3.56694986e-03 8.25459406e-12 9.96433020e-01]\n",
"2.0 [ 8.97541104e-05 1.74763748e-09 9.99910235e-01]\n",
"2.0 [ 3.33245983e-03 5.36455082e-16 9.96667564e-01]\n",
"2.0 [ 2.65802100e-07 1.98834368e-06 9.99997735e-01]\n",
"2.0 [ 1.29129319e-06 2.07540140e-15 9.99998689e-01]\n",
"2.0 [ 2.08257589e-09 4.46392532e-12 1.00000000e+00]\n",
"2.0 [ 3.98147131e-05 8.14328871e-09 9.99960184e-01]\n",
"2.0 [ 5.01298600e-05 1.04539940e-05 9.99939442e-01]\n",
"2.0 [ 5.25748124e-04 7.26537937e-15 9.99474227e-01]\n",
"2.0 [ 2.35757497e-07 6.30105551e-11 9.99999762e-01]\n",
"2.0 [ 1.04627409e-03 5.31124655e-08 9.98953700e-01]\n",
"2.0 [ 3.62033909e-03 2.82681071e-07 9.96379435e-01]\n",
"2.0 [ 2.03396613e-03 4.33538116e-06 9.97961760e-01]\n",
"2.0 [ 2.53706914e-03 2.81509529e-06 9.97460186e-01]\n",
"2.0 [ 2.29555159e-03 1.18074684e-04 9.97586370e-01]\n",
"2.0 [ 3.17676324e-07 3.52383613e-05 9.99964476e-01]\n"
]
},
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAksAAAJCCAYAAADQsoPKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQ1gCAOgQlgAAOv4/rJ1YohnTwAUAAAAASUVORK5CYII=\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x7f63c692d390>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"# predict the same image again\n",
"classification, labels, cls_loss, anchors = model.predict_on_batch(inputs)\n",
"\n",
"# visualize the positive anchors\n",
"indices = np.where(labels[0, :] > 0)[0]\n",
"im = image_batch[0, ...].copy()\n",
"print(labels.shape, classification.shape, anchors.shape)\n",
"for idx in indices:\n",
" # print the label and the classification scores\n",
" print(labels[0, idx], classification[0, idx, :])\n",
" b = anchors[0, idx, :].astype(int)\n",
" cv2.rectangle(im, (b[0], b[1]), (b[2], b[3]), (0, 1, 0), 3)\n",
" \n",
"plt.figure(figsize=(10, 10))\n",
"plt.imshow(im)\n",
"plt.show()"
]
}
],
"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": "3.6.2"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment