/home/ashok/anaconda3/bin/python /home/ashok/PycharmProjects/Tensorflow/food-image-classification--master/imageClassificationByCNN.py Using TensorFlow backend. Found 4654 images belonging to 10 classes. Found 1168 images belonging to 10 classes. /home/ashok/PycharmProjects/Tensorflow/food-image-classification--master/imageClassificationByCNN.py:121: UserWarning: Update your Conv2D call to the Keras 2 API: Conv2D(32, (7, 7), input_shape=(128, 128,...) model.add(Convolution2D(32, 7, 7, input_shape=(128, 128, 3))) /home/ashok/PycharmProjects/Tensorflow/food-image-classification--master/imageClassificationByCNN.py:124: UserWarning: Update your Conv2D call to the Keras 2 API: Conv2D(64, (5, 5)) model.add(Convolution2D(64, 5, 5)) /home/ashok/PycharmProjects/Tensorflow/food-image-classification--master/imageClassificationByCNN.py:127: UserWarning: Update your Conv2D call to the Keras 2 API: Conv2D(128, (3, 3)) model.add(Convolution2D(128, 3, 3))

Layer (type) Output Shape Param #

conv2d_1 (Conv2D) (None, 122, 122, 32) 4736

activation_1 (Activation) (None, 122, 122, 32) 0

max_pooling2d_1 (MaxPooling2 (None, 61, 61, 32) 0

conv2d_2 (Conv2D) (None, 57, 57, 64) 51264

activation_2 (Activation) (None, 57, 57, 64) 0

max_pooling2d_2 (MaxPooling2 (None, 28, 28, 64) 0

conv2d_3 (Conv2D) (None, 26, 26, 128) 73856

activation_3 (Activation) (None, 26, 26, 128) 0

max_pooling2d_3 (MaxPooling2 (None, 13, 13, 128) 0

dropout_1 (Dropout) (None, 13, 13, 128) 0

flatten_1 (Flatten) (None, 21632) 0

dense_1 (Dense) (None, 128) 2769024

activation_4 (Activation) (None, 128) 0

dropout_2 (Dropout) (None, 128) 0

dense_2 (Dense) (None, 10) 1290

activation_5 (Activation) (None, 10) 0

Total params: 2,900,170.0 Trainable params: 2,900,170.0 Non-trainable params: 0.0

None /home/ashok/PycharmProjects/Tensorflow/food-image-classification--master/imageClassificationByCNN.py:188: UserWarning: Update your fit_generator call to the Keras 2 API: fit_generator(<keras.pre..., verbose=1, validation_data=<keras.pre..., callbacks=[<keras.ca..., steps_per_epoch=145, epochs=100, validation_steps=1168) callbacks = [earlystopper, lrate, checkpoint, hist]) Epoch 1/100 2017-07-31 10:22:34.909867: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.1 instructions, but these are available on your machine and could speed up CPU computations. 2017-07-31 10:22:34.909906: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations. 2017-07-31 10:22:34.909917: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations. 2017-07-31 10:22:34.909925: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX2 instructions, but these are available on your machine and could speed up CPU computations. 2017-07-31 10:22:34.909933: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use FMA instructions, but these are available on your machine and could speed up CPU computations. 1/145 [..............................] - ETA: 858s - loss: 2.3407 - acc: 0.0625 2/145 [..............................] - ETA: 686s - loss: 2.3314 - acc: 0.1094 3/145 [..............................] - ETA: 619s - loss: 2.3092 - acc: 0.1250 4/145 [..............................] - ETA: 583s - loss: 2.3028 - acc: 0.1562 5/145 [>.............................] - ETA: 560s - loss: 2.2906 - acc: 0.1875 6/145 [>.............................] - ETA: 544s - loss: 2.2735 - acc: 0.2031 7/145 [>.............................] - ETA: 532s - loss: 2.2548 - acc: 0.2098 8/145 [>.............................] - ETA: 521s - loss: 2.2769 - acc: 0.2070 9/145 [>.............................] - ETA: 512s - loss: 2.2668 - acc: 0.2188 10/145 [=>............................] - ETA: 503s - loss: 2.2576 - acc: 0.2188 11/145 [=>............................] - ETA: 496s - loss: 2.2542 - acc: 0.2131 12/145 [=>............................] - ETA: 489s - loss: 2.2457 - acc: 0.2135 13/145 [=>............................] - ETA: 483s - loss: 2.2414 - acc: 0.2163 14/145 [=>............................] - ETA: 477s - loss: 2.2427 - acc: 0.2143 15/145 [==>...........................] - ETA: 472s - loss: 2.2332 - acc: 0.2146 16/145 [==>...........................] - ETA: 466s - loss: 2.2248 - acc: 0.2148 17/145 [==>...........................] - ETA: 461s - loss: 2.2325 - acc: 0.2151 18/145 [==>...........................] - ETA: 456s - loss: 2.2320 - acc: 0.2135 19/145 [==>...........................] - ETA: 452s - loss: 2.2321 - acc: 0.2171 20/145 [===>..........................] - ETA: 447s - loss: 2.2338 - acc: 0.2156 21/145 [===>..........................] - ETA: 444s - loss: 2.2291 - acc: 0.2158 22/145 [===>..........................] - ETA: 440s - loss: 2.2230 - acc: 0.2159 23/145 [===>..........................] - ETA: 435s - loss: 2.2234 - acc: 0.2160 24/145 [===>..........................] - ETA: 431s - loss: 2.2170 - acc: 0.2201 25/145 [====>.........................] - ETA: 427s - loss: 2.2212 - acc: 0.2150 26/145 [====>.........................] - ETA: 423s - loss: 2.2185 - acc: 0.2188 27/145 [====>.........................] - ETA: 419s - loss: 2.2147 - acc: 0.2199 28/145 [====>.........................] - ETA: 417s - loss: 2.2144 - acc: 0.2176 29/145 [=====>........................] - ETA: 415s - loss: 2.2071 - acc: 0.2166 30/145 [=====>........................] - ETA: 412s - loss: 2.1974 - acc: 0.2177 31/145 [=====>........................] - ETA: 409s - loss: 2.1981 - acc: 0.2147 32/145 [=====>........................] - ETA: 406s - loss: 2.1949 - acc: 0.2148 33/145 [=====>........................] - ETA: 403s - loss: 2.1962 - acc: 0.2140 34/145 [======>.......................] - ETA: 399s - loss: 2.1949 - acc: 0.2142 35/145 [======>.......................] - ETA: 397s - loss: 2.1944 - acc: 0.2134 36/145 [======>.......................] - ETA: 394s - loss: 2.1894 - acc: 0.2153 37/145 [======>.......................] - ETA: 392s - loss: 2.1966 - acc: 0.2162 38/145 [======>.......................] - ETA: 391s - loss: 2.1942 - acc: 0.2171 39/145 [=======>......................] - ETA: 387s - loss: 2.1928 - acc: 0.2163 40/145 [=======>......................] - ETA: 384s - loss: 2.1878 - acc: 0.2180 41/145 [=======>......................] - ETA: 380s - loss: 2.1854 - acc: 0.2203 42/145 [=======>......................] - ETA: 377s - loss: 2.1826 - acc: 0.2202 43/145 [=======>......................] - ETA: 375s - loss: 2.1807 - acc: 0.2195 44/145 [========>.....................] - ETA: 373s - loss: 2.1782 - acc: 0.2202 45/145 [========>.....................] - ETA: 370s - loss: 2.1766 - acc: 0.2222 46/145 [========>.....................] - ETA: 367s - loss: 2.1713 - acc: 0.2262 47/145 [========>.....................] - ETA: 365s - loss: 2.1703 - acc: 0.2267 48/145 [========>.....................] - ETA: 362s - loss: 2.1696 - acc: 0.2259 49/145 [=========>....................] - ETA: 358s - loss: 2.1673 - acc: 0.2283 50/145 [=========>....................] - ETA: 354s - loss: 2.1631 - acc: 0.2281 51/145 [=========>....................] - ETA: 351s - loss: 2.1585 - acc: 0.2273 52/145 [=========>....................] - ETA: 347s - loss: 2.1520 - acc: 0.2302 53/145 [=========>....................] - ETA: 343s - loss: 2.1488 - acc: 0.2300 54/145 [==========>...................] - ETA: 339s - loss: 2.1460 - acc: 0.2309 55/145 [==========>...................] - ETA: 334s - loss: 2.1435 - acc: 0.2307 56/145 [==========>...................] - ETA: 330s - loss: 2.1406 - acc: 0.2321 57/145 [==========>...................] - ETA: 326s - loss: 2.1392 - acc: 0.2319 58/145 [===========>..................] - ETA: 322s - loss: 2.1359 - acc: 0.2322 59/145 [===========>..................] - ETA: 318s - loss: 2.1303 - acc: 0.2336 60/145 [===========>..................] - ETA: 314s - loss: 2.1275 - acc: 0.2344 61/145 [===========>..................] - ETA: 310s - loss: 2.1252 - acc: 0.2346 62/145 [===========>..................] - ETA: 306s - loss: 2.1219 - acc: 0.2364 63/145 [============>.................] - ETA: 302s - loss: 2.1173 - acc: 0.2376 64/145 [============>.................] - ETA: 298s - loss: 2.1150 - acc: 0.2388 65/145 [============>.................] - ETA: 294s - loss: 2.1105 - acc: 0.2413 66/145 [============>.................] - ETA: 290s - loss: 2.1109 - acc: 0.2401 67/145 [============>.................] - ETA: 286s - loss: 2.1071 - acc: 0.2416 68/145 [=============>................] - ETA: 282s - loss: 2.1023 - acc: 0.2422 69/145 [=============>................] - ETA: 278s - loss: 2.0987 - acc: 0.2423 70/145 [=============>................] - ETA: 274s - loss: 2.1017 - acc: 0.2429 71/145 [=============>................] - ETA: 270s - loss: 2.1047 - acc: 0.2408 72/145 [=============>................] - ETA: 266s - loss: 2.1044 - acc: 0.2413 73/145 [==============>...............] - ETA: 262s - loss: 2.1038 - acc: 0.2423 74/145 [==============>...............] - ETA: 259s - loss: 2.1027 - acc: 0.2428 75/145 [==============>...............] - ETA: 255s - loss: 2.1017 - acc: 0.2433 76/145 [==============>...............] - ETA: 251s - loss: 2.1014 - acc: 0.2426 77/145 [==============>...............] - ETA: 247s - loss: 2.0975 - acc: 0.2447 78/145 [===============>..............] - ETA: 243s - loss: 2.0960 - acc: 0.2452 79/145 [===============>..............] - ETA: 239s - loss: 2.0941 - acc: 0.2460 80/145 [===============>..............] - ETA: 235s - loss: 2.0937 - acc: 0.2465 81/145 [===============>..............] - ETA: 232s - loss: 2.0931 - acc: 0.2450 82/145 [===============>..............] - ETA: 228s - loss: 2.0927 - acc: 0.2454 83/145 [================>.............] - ETA: 224s - loss: 2.0919 - acc: 0.2451 84/145 [================>.............] - ETA: 220s - loss: 2.0921 - acc: 0.2463 85/145 [================>.............] - ETA: 217s - loss: 2.0889 - acc: 0.2467 86/145 [================>.............] - ETA: 213s - loss: 2.0857 - acc: 0.2485 87/145 [=================>............] - ETA: 209s - loss: 2.0847 - acc: 0.2486 88/145 [=================>............] - ETA: 205s - loss: 2.0823 - acc: 0.2493 89/145 [=================>............] - ETA: 202s - loss: 2.0827 - acc: 0.2489 90/145 [=================>............] - ETA: 198s - loss: 2.0819 - acc: 0.2490 91/145 [=================>............] - ETA: 194s - loss: 2.0819 - acc: 0.2483 92/145 [==================>...........] - ETA: 190s - loss: 2.0781 - acc: 0.2486 93/145 [==================>...........] - ETA: 187s - loss: 2.0758 - acc: 0.2487 94/145 [==================>...........] - ETA: 183s - loss: 2.0728 - acc: 0.2500 95/145 [==================>...........] - ETA: 179s - loss: 2.0721 - acc: 0.2510 96/145 [==================>...........] - ETA: 176s - loss: 2.0676 - acc: 0.2523 97/145 [===================>..........] - ETA: 172s - loss: 2.0625 - acc: 0.2545 98/145 [===================>..........] - ETA: 168s - loss: 2.0613 - acc: 0.2554 99/145 [===================>..........] - ETA: 165s - loss: 2.0572 - acc: 0.2563 100/145 [===================>..........] - ETA: 161s - loss: 2.0519 - acc: 0.2578 101/145 [===================>..........] - ETA: 157s - loss: 2.0457 - acc: 0.2605 102/145 [====================>.........] - ETA: 154s - loss: 2.0494 - acc: 0.2607 103/145 [====================>.........] - ETA: 150s - loss: 2.0489 - acc: 0.2615 104/145 [====================>.........] - ETA: 146s - loss: 2.0451 - acc: 0.2632 105/145 [====================>.........] - ETA: 143s - loss: 2.0438 - acc: 0.2640 106/145 [====================>.........] - ETA: 139s - loss: 2.0407 - acc: 0.2647 107/145 [=====================>........] - ETA: 135s - loss: 2.0384 - acc: 0.2658 108/145 [=====================>........] - ETA: 132s - loss: 2.0368 - acc: 0.2674 109/145 [=====================>........] - ETA: 128s - loss: 2.0336 - acc: 0.2683 110/145 [=====================>........] - ETA: 125s - loss: 2.0334 - acc: 0.2693 111/145 [=====================>........] - ETA: 121s - loss: 2.0322 - acc: 0.2706 112/145 [======================>.......] - ETA: 117s - loss: 2.0315 - acc: 0.2712 113/145 [======================>.......] - ETA: 114s - loss: 2.0290 - acc: 0.2721 114/145 [======================>.......] - ETA: 110s - loss: 2.0298 - acc: 0.2728 115/145 [======================>.......] - ETA: 106s - loss: 2.0299 - acc: 0.2717 116/145 [=======================>......] - ETA: 103s - loss: 2.0273 - acc: 0.2724 117/145 [=======================>......] - ETA: 99s - loss: 2.0264 - acc: 0.2727 118/145 [=======================>......] - ETA: 96s - loss: 2.0261 - acc: 0.2728 119/145 [=======================>......] - ETA: 92s - loss: 2.0264 - acc: 0.2731 120/145 [=======================>......] - ETA: 88s - loss: 2.0255 - acc: 0.2732 121/145 [========================>.....] - ETA: 85s - loss: 2.0238 - acc: 0.2740 122/145 [========================>.....] - ETA: 81s - loss: 2.0202 - acc: 0.2759 123/145 [========================>.....] - ETA: 78s - loss: 2.0176 - acc: 0.2764 124/145 [========================>.....] - ETA: 74s - loss: 2.0171 - acc: 0.2760 125/145 [========================>.....] - ETA: 71s - loss: 2.0151 - acc: 0.2762 126/145 [=========================>....] - ETA: 67s - loss: 2.0134 - acc: 0.2773 127/145 [=========================>....] - ETA: 63s - loss: 2.0111 - acc: 0.2785 128/145 [=========================>....] - ETA: 60s - loss: 2.0086 - acc: 0.2791 129/145 [=========================>....] - ETA: 56s - loss: 2.0091 - acc: 0.2791 130/145 [=========================>....] - ETA: 53s - loss: 2.0077 - acc: 0.2793 131/145 [==========================>...] - ETA: 49s - loss: 2.0094 - acc: 0.2793 132/145 [==========================>...] - ETA: 46s - loss: 2.0085 - acc: 0.2791 133/145 [==========================>...] - ETA: 42s - loss: 2.0080 - acc: 0.2798 134/145 [==========================>...] - ETA: 38s - loss: 2.0063 - acc: 0.2796 135/145 [==========================>...] - ETA: 35s - loss: 2.0054 - acc: 0.2796 136/145 [===========================>..] - ETA: 31s - loss: 2.0047 - acc: 0.2796 137/145 [===========================>..] - ETA: 28s - loss: 2.0032 - acc: 0.2801 138/145 [===========================>..] - ETA: 24s - loss: 2.0050 - acc: 0.2799 139/145 [===========================>..] - ETA: 21s - loss: 2.0051 - acc: 0.2806 140/145 [===========================>..] - ETA: 17s - loss: 2.0035 - acc: 0.2815 141/145 [============================>.] - ETA: 14s - loss: 2.0037 - acc: 0.2812 142/145 [============================>.] - ETA: 10s - loss: 2.0018 - acc: 0.2819 143/145 [============================>.] - ETA: 7s - loss: 2.0009 - acc: 0.2815 144/145 [============================>.] - ETA: 3s - loss: 1.9996 - acc: 0.2823loss: 1.99745497539 New learning rate: 0.00737027467905 Epoch 00000: val_acc improved from -inf to 0.31695, saving model to /home/ashok/Desktop/out/modelWeights/cnnModelDEp80weights.best.hdf5 Traceback (most recent call last): File "/home/ashok/PycharmProjects/Tensorflow/food-image-classification--master/imageClassificationByCNN.py", line 188, in callbacks = [earlystopper, lrate, checkpoint, hist]) File "/home/ashok/anaconda3/lib/python3.6/site-packages/keras/legacy/interfaces.py", line 88, in wrapper return func(*args, *kwargs) File "/home/ashok/anaconda3/lib/python3.6/site-packages/keras/models.py", line 1097, in fit_generator initial_epoch=initial_epoch) File "/home/ashok/anaconda3/lib/python3.6/site-packages/keras/legacy/interfaces.py", line 88, in wrapper return func(args, **kwargs) File "/home/ashok/anaconda3/lib/python3.6/site-packages/keras/engine/training.py", line 1913, in fit_generator callbacks.on_epoch_end(epoch, epoch_logs) File "/home/ashok/anaconda3/lib/python3.6/site-packages/keras/callbacks.py", line 75, in on_epoch_end callback.on_epoch_end(epoch, logs) File "/home/ashok/anaconda3/lib/python3.6/site-packages/keras/callbacks.py", line 400, in on_epoch_end self.model.save(filepath, overwrite=True) File "/home/ashok/anaconda3/lib/python3.6/site-packages/keras/engine/topology.py", line 2429, in save save_model(self, filepath, overwrite) File "/home/ashok/anaconda3/lib/python3.6/site-packages/keras/models.py", line 96, in save_model f = h5py.File(filepath, 'w') File "/home/ashok/anaconda3/lib/python3.6/site-packages/h5py/_hl/files.py", line 271, in init fid = make_fid(name, mode, userblock_size, fapl, swmr=swmr) File "/home/ashok/anaconda3/lib/python3.6/site-packages/h5py/_hl/files.py", line 107, in make_fid fid = h5f.create(name, h5f.ACC_TRUNC, fapl=fapl, fcpl=fcpl) File "h5py/_objects.pyx", line 54, in h5py._objects.with_phil.wrapper (/home/ilan/minonda/conda-bld/h5py_1490028290543/work/h5py/_objects.c:2846) File "h5py/_objects.pyx", line 55, in h5py._objects.with_phil.wrapper (/home/ilan/minonda/conda-bld/h5py_1490028290543/work/h5py/_objects.c:2804) File "h5py/h5f.pyx", line 98, in h5py.h5f.create (/home/ilan/minonda/conda-bld/h5py_1490028290543/work/h5py/h5f.c:2290) OSError: Unable to create file (Unable to open file: name = '/home/ashok/desktop/out/modelweights/cnnmodeldep80weights.best.hdf5', errno = 2, error message = 'no such file or directory', flags = 13, o_flags = 242)

Process finished with exit code 1

Please note the original code is based on keras 1.1.0 and theano 0.82. If you use tensorflow, I cannot guarantee if it works or not

Thank you for your reply but the thing is that the same error is coming again even the keras with back end theano

/usr/bin/python2.7 /home/ashok/PycharmProjects/Tensorflow/food-image-classification--master/imageClassificationByCNN.py Using Theano backend. /home/ashok/.local/lib/python2.7/site-packages/keras/backend/theano_backend.py:1282: UserWarning: DEPRECATION: the 'ds' parameter is not going to exist anymore as it is going to be replaced by the parameter 'ws'. mode='max') /home/ashok/.local/lib/python2.7/site-packages/keras/backend/theano_backend.py:1282: UserWarning: DEPRECATION: the 'st' parameter is not going to exist anymore as it is going to be replaced by the parameter 'stride'. mode='max') /home/ashok/.local/lib/python2.7/site-packages/keras/backend/theano_backend.py:1282: UserWarning: DEPRECATION: the 'padding' parameter is not going to exist anymore as it is going to be replaced by the parameter 'pad'. mode='max')

Layer (type) Output Shape Param # Connected to

convolution2d_1 (Convolution2D) (None, 122, 122, 32) 4736 convolution2d_input_1[0][0]

activation_1 (Activation) (None, 122, 122, 32) 0 convolution2d_1[0][0]

maxpooling2d_1 (MaxPooling2D) (None, 61, 61, 32) 0 activation_1[0][0]

convolution2d_2 (Convolution2D) (None, 57, 57, 64) 51264 maxpooling2d_1[0][0]

activation_2 (Activation) (None, 57, 57, 64) 0 convolution2d_2[0][0]

maxpooling2d_2 (MaxPooling2D) (None, 28, 28, 64) 0 activation_2[0][0]

convolution2d_3 (Convolution2D) (None, 26, 26, 128) 73856 maxpooling2d_2[0][0]

activation_3 (Activation) (None, 26, 26, 128) 0 convolution2d_3[0][0]

maxpooling2d_3 (MaxPooling2D) (None, 13, 13, 128) 0 activation_3[0][0]

dropout_1 (Dropout) (None, 13, 13, 128) 0 maxpooling2d_3[0][0]

flatten_1 (Flatten) (None, 21632) 0 dropout_1[0][0]

dense_1 (Dense) (None, 128) 2769024 flatten_1[0][0]

activation_4 (Activation) (None, 128) 0 dense_1[0][0]

dropout_2 (Dropout) (None, 128) 0 activation_4[0][0]

dense_2 (Dense) (None, 10) 1290 dropout_2[0][0]

activation_5 (Activation) (None, 10) 0 dense_2[0][0]

Total params: 2900170

None ('hello', '/home/ashok/Desktop/out/modelWeights/cnnModelDEp80weights.best.hdf5') Found 4654 images belonging to 10 classes. Found 1168 images belonging to 10 classes. Epoch 1/1 32/4654 [..............................] - ETA: 577s - loss: 2.3012 - acc: 0.0312 64/4654 [..............................] - ETA: 579s - loss: 2.3018 - acc: 0.0938 96/4654 [..............................] - ETA: 578s - loss: 2.3013 - acc: 0.0833 128/4654 [..............................] - ETA: 578s - loss: 2.3004 - acc: 0.0938 160/4654 [>.............................] - ETA: 572s - loss: 2.2989 - acc: 0.1125 192/4654 [>.............................] - ETA: 568s - loss: 2.2959 - acc: 0.1094 224/4654 [>.............................] - ETA: 565s - loss: 2.2935 - acc: 0.1161 256/4654 [>.............................] - ETA: 561s - loss: 2.2915 - acc: 0.1211 288/4654 [>.............................] - ETA: 560s - loss: 2.2863 - acc: 0.1389 320/4654 [=>............................] - ETA: 556s - loss: 2.2822 - acc: 0.1469 352/4654 [=>............................] - ETA: 552s - loss: 2.2768 - acc: 0.1449 384/4654 [=>............................] - ETA: 547s - loss: 2.2658 - acc: 0.1562 416/4654 [=>............................] - ETA: 544s - loss: 2.2711 - acc: 0.1611 448/4654 [=>............................] - ETA: 540s - loss: 2.2673 - acc: 0.1674 480/4654 [==>...........................] - ETA: 535s - loss: 2.2703 - acc: 0.1667 512/4654 [==>...........................] - ETA: 530s - loss: 2.2702 - acc: 0.1582 544/4654 [==>...........................] - ETA: 526s - loss: 2.2635 - acc: 0.1654 576/4654 [==>...........................] - ETA: 522s - loss: 2.2667 - acc: 0.1615 608/4654 [==>...........................] - ETA: 518s - loss: 2.2613 - acc: 0.1628 640/4654 [===>..........................] - ETA: 514s - loss: 2.2563 - acc: 0.1672 672/4654 [===>..........................] - ETA: 510s - loss: 2.2580 - acc: 0.1637 704/4654 [===>..........................] - ETA: 506s - loss: 2.2568 - acc: 0.1634 736/4654 [===>..........................] - ETA: 502s - loss: 2.2581 - acc: 0.1630 768/4654 [===>..........................] - ETA: 497s - loss: 2.2534 - acc: 0.1667 800/4654 [====>.........................] - ETA: 493s - loss: 2.2496 - acc: 0.1737 832/4654 [====>.........................] - ETA: 491s - loss: 2.2485 - acc: 0.1731 864/4654 [====>.........................] - ETA: 487s - loss: 2.2421 - acc: 0.1771 896/4654 [====>.........................] - ETA: 484s - loss: 2.2347 - acc: 0.1808 928/4654 [====>.........................] - ETA: 481s - loss: 2.2303 - acc: 0.1853 960/4654 [=====>........................] - ETA: 477s - loss: 2.2299 - acc: 0.1865 992/4654 [=====>........................] - ETA: 472s - loss: 2.2283 - acc: 0.1855 1024/4654 [=====>........................] - ETA: 468s - loss: 2.2254 - acc: 0.1885 1056/4654 [=====>........................] - ETA: 464s - loss: 2.2237 - acc: 0.1884 1088/4654 [======>.......................] - ETA: 461s - loss: 2.2260 - acc: 0.1866 1120/4654 [======>.......................] - ETA: 457s - loss: 2.2163 - acc: 0.1955 1152/4654 [======>.......................] - ETA: 453s - loss: 2.2145 - acc: 0.1988 1184/4654 [======>.......................] - ETA: 449s - loss: 2.2093 - acc: 0.2010 1216/4654 [======>.......................] - ETA: 444s - loss: 2.2038 - acc: 0.2023 1248/4654 [=======>......................] - ETA: 440s - loss: 2.2048 - acc: 0.2027 1280/4654 [=======>......................] - ETA: 435s - loss: 2.2046 - acc: 0.2031 1312/4654 [=======>......................] - ETA: 431s - loss: 2.2048 - acc: 0.2012 1344/4654 [=======>......................] - ETA: 427s - loss: 2.2053 - acc: 0.2001 1376/4654 [=======>......................] - ETA: 422s - loss: 2.2043 - acc: 0.2013 1408/4654 [========>.....................] - ETA: 418s - loss: 2.2044 - acc: 0.2031 1440/4654 [========>.....................] - ETA: 413s - loss: 2.2039 - acc: 0.2028 1472/4654 [========>.....................] - ETA: 409s - loss: 2.2028 - acc: 0.2038 1504/4654 [========>.....................] - ETA: 405s - loss: 2.2012 - acc: 0.2028 1536/4654 [========>.....................] - ETA: 400s - loss: 2.2001 - acc: 0.2025 1568/4654 [=========>....................] - ETA: 396s - loss: 2.2009 - acc: 0.2015 1600/4654 [=========>....................] - ETA: 392s - loss: 2.2008 - acc: 0.2013 1632/4654 [=========>....................] - ETA: 388s - loss: 2.1998 - acc: 0.2034 1664/4654 [=========>....................] - ETA: 383s - loss: 2.1991 - acc: 0.2025 1696/4654 [=========>....................] - ETA: 379s - loss: 2.1981 - acc: 0.2017 1728/4654 [==========>...................] - ETA: 375s - loss: 2.1982 - acc: 0.2014 1760/4654 [==========>...................] - ETA: 371s - loss: 2.1966 - acc: 0.2023 1792/4654 [==========>...................] - ETA: 366s - loss: 2.1948 - acc: 0.2037 1824/4654 [==========>...................] - ETA: 362s - loss: 2.1955 - acc: 0.2023 1856/4654 [==========>...................] - ETA: 358s - loss: 2.1977 - acc: 0.2031 1888/4654 [===========>..................] - ETA: 354s - loss: 2.1989 - acc: 0.2018 1920/4654 [===========>..................] - ETA: 349s - loss: 2.1974 - acc: 0.2026 1952/4654 [===========>..................] - ETA: 345s - loss: 2.1966 - acc: 0.2034 1984/4654 [===========>..................] - ETA: 341s - loss: 2.1949 - acc: 0.2041 2016/4654 [===========>..................] - ETA: 337s - loss: 2.1909 - acc: 0.2034 2048/4654 [============>.................] - ETA: 332s - loss: 2.1899 - acc: 0.2036 2080/4654 [============>.................] - ETA: 328s - loss: 2.1922 - acc: 0.2038 2112/4654 [============>.................] - ETA: 324s - loss: 2.1901 - acc: 0.2045 2144/4654 [============>.................] - ETA: 320s - loss: 2.1904 - acc: 0.2043 2176/4654 [=============>................] - ETA: 316s - loss: 2.1898 - acc: 0.2036 2208/4654 [=============>................] - ETA: 312s - loss: 2.1901 - acc: 0.2038 2240/4654 [=============>................] - ETA: 307s - loss: 2.1890 - acc: 0.2036 2272/4654 [=============>................] - ETA: 303s - loss: 2.1891 - acc: 0.2038 2304/4654 [=============>................] - ETA: 299s - loss: 2.1864 - acc: 0.2049 2336/4654 [==============>...............] - ETA: 295s - loss: 2.1854 - acc: 0.2046 2368/4654 [==============>...............] - ETA: 291s - loss: 2.1867 - acc: 0.2048 2400/4654 [==============>...............] - ETA: 287s - loss: 2.1858 - acc: 0.2050 2432/4654 [==============>...............] - ETA: 283s - loss: 2.1864 - acc: 0.2039 2464/4654 [==============>...............] - ETA: 279s - loss: 2.1873 - acc: 0.2033 2496/4654 [===============>..............] - ETA: 275s - loss: 2.1866 - acc: 0.2043 2528/4654 [===============>..............] - ETA: 270s - loss: 2.1857 - acc: 0.2033 2560/4654 [===============>..............] - ETA: 266s - loss: 2.1866 - acc: 0.2035 2592/4654 [===============>..............] - ETA: 262s - loss: 2.1855 - acc: 0.2033 2624/4654 [===============>..............] - ETA: 258s - loss: 2.1854 - acc: 0.2024 2656/4654 [================>.............] - ETA: 254s - loss: 2.1843 - acc: 0.2029 2688/4654 [================>.............] - ETA: 250s - loss: 2.1837 - acc: 0.2035 2720/4654 [================>.............] - ETA: 246s - loss: 2.1814 - acc: 0.2033 2752/4654 [================>.............] - ETA: 242s - loss: 2.1800 - acc: 0.2035 2784/4654 [================>.............] - ETA: 237s - loss: 2.1796 - acc: 0.2029 2816/4654 [=================>............] - ETA: 233s - loss: 2.1787 - acc: 0.2031 2848/4654 [=================>............] - ETA: 229s - loss: 2.1779 - acc: 0.2026 2880/4654 [=================>............] - ETA: 225s - loss: 2.1756 - acc: 0.2031 2912/4654 [=================>............] - ETA: 221s - loss: 2.1741 - acc: 0.2023 2944/4654 [=================>............] - ETA: 217s - loss: 2.1716 - acc: 0.2028 2976/4654 [==================>...........] - ETA: 213s - loss: 2.1703 - acc: 0.2026 3008/4654 [==================>...........] - ETA: 209s - loss: 2.1702 - acc: 0.2021 3040/4654 [==================>...........] - ETA: 205s - loss: 2.1695 - acc: 0.2023 3072/4654 [==================>...........] - ETA: 200s - loss: 2.1670 - acc: 0.2038 3104/4654 [===================>..........] - ETA: 196s - loss: 2.1659 - acc: 0.2030 3136/4654 [===================>..........] - ETA: 192s - loss: 2.1639 - acc: 0.2031 3168/4654 [===================>..........] - ETA: 188s - loss: 2.1615 - acc: 0.2039 3200/4654 [===================>..........] - ETA: 184s - loss: 2.1607 - acc: 0.2037 3232/4654 [===================>..........] - ETA: 180s - loss: 2.1610 - acc: 0.2039 3264/4654 [====================>.........] - ETA: 176s - loss: 2.1602 - acc: 0.2047 3296/4654 [====================>.........] - ETA: 172s - loss: 2.1601 - acc: 0.2051 3328/4654 [====================>.........] - ETA: 168s - loss: 2.1584 - acc: 0.2067 3360/4654 [====================>.........] - ETA: 164s - loss: 2.1572 - acc: 0.2074 3392/4654 [====================>.........] - ETA: 160s - loss: 2.1561 - acc: 0.2081 3424/4654 [=====================>........] - ETA: 155s - loss: 2.1538 - acc: 0.2097 3456/4654 [=====================>........] - ETA: 151s - loss: 2.1526 - acc: 0.2101 3488/4654 [=====================>........] - ETA: 147s - loss: 2.1510 - acc: 0.2113 3520/4654 [=====================>........] - ETA: 143s - loss: 2.1522 - acc: 0.2111 3552/4654 [=====================>........] - ETA: 139s - loss: 2.1505 - acc: 0.2123 3584/4654 [======================>.......] - ETA: 135s - loss: 2.1474 - acc: 0.2137 3616/4654 [======================>.......] - ETA: 131s - loss: 2.1447 - acc: 0.2140 3648/4654 [======================>.......] - ETA: 127s - loss: 2.1440 - acc: 0.2144 3680/4654 [======================>.......] - ETA: 123s - loss: 2.1423 - acc: 0.2144 3712/4654 [======================>.......] - ETA: 119s - loss: 2.1420 - acc: 0.2142 3744/4654 [=======================>......] - ETA: 115s - loss: 2.1412 - acc: 0.2145 3776/4654 [=======================>......] - ETA: 111s - loss: 2.1406 - acc: 0.2145 3808/4654 [=======================>......] - ETA: 107s - loss: 2.1381 - acc: 0.2159 3840/4654 [=======================>......] - ETA: 103s - loss: 2.1342 - acc: 0.2177 3872/4654 [=======================>......] - ETA: 99s - loss: 2.1337 - acc: 0.2169 3904/4654 [========================>.....] - ETA: 95s - loss: 2.1322 - acc: 0.2164 3936/4654 [========================>.....] - ETA: 90s - loss: 2.1312 - acc: 0.2157 3968/4654 [========================>.....] - ETA: 86s - loss: 2.1305 - acc: 0.2157 4000/4654 [========================>.....] - ETA: 82s - loss: 2.1285 - acc: 0.2162 4032/4654 [========================>.....] - ETA: 78s - loss: 2.1275 - acc: 0.2163 4064/4654 [=========================>....] - ETA: 74s - loss: 2.1252 - acc: 0.2180 4096/4654 [=========================>....] - ETA: 70s - loss: 2.1235 - acc: 0.2195 4128/4654 [=========================>....] - ETA: 66s - loss: 2.1219 - acc: 0.2192 4160/4654 [=========================>....] - ETA: 62s - loss: 2.1226 - acc: 0.2188 4192/4654 [==========================>...] - ETA: 58s - loss: 2.1203 - acc: 0.2197 4224/4654 [==========================>...] - ETA: 54s - loss: 2.1178 - acc: 0.2206 4256/4654 [==========================>...] - ETA: 50s - loss: 2.1162 - acc: 0.2216 4288/4654 [==========================>...] - ETA: 46s - loss: 2.1149 - acc: 0.2215 4320/4654 [==========================>...] - ETA: 42s - loss: 2.1130 - acc: 0.2222 4352/4654 [===========================>..] - ETA: 38s - loss: 2.1119 - acc: 0.2231 4384/4654 [===========================>..] - ETA: 34s - loss: 2.1127 - acc: 0.2231 4416/4654 [===========================>..] - ETA: 30s - loss: 2.1115 - acc: 0.2231 4448/4654 [===========================>..] - ETA: 26s - loss: 2.1101 - acc: 0.2230 4480/4654 [===========================>..] - ETA: 22s - loss: 2.1083 - acc: 0.2232 4512/4654 [============================>.] - ETA: 17s - loss: 2.1066 - acc: 0.2238 4544/4654 [============================>.] - ETA: 13s - loss: 2.1054 - acc: 0.2245 4576/4654 [============================>.] - ETA: 9s - loss: 2.1043 - acc: 0.2247 4608/4654 [============================>.] - ETA: 5s - loss: 2.1037 - acc: 0.2250 4640/4654 [============================>.] - ETA: 1s - loss: 2.1020 - acc: 0.2259('loss: ', 2.1008654166866281) ('New learning rate: ', 0.0081732401111684842) Epoch 00000: val_acc improved from -inf to 0.32192, saving model to /home/ashok/Desktop/out/modelWeights/cnnModelDEp80weights.best.hdf5 Traceback (most recent call last): File "/home/ashok/PycharmProjects/Tensorflow/food-image-classification--master/imageClassificationByCNN.py", line 190, in callbacks = [earlystopper, lrate, checkpoint, hist]) File "/home/ashok/.local/lib/python2.7/site-packages/keras/models.py", line 874, in fit_generator pickle_safe=pickle_safe) File "/home/ashok/.local/lib/python2.7/site-packages/keras/engine/training.py", line 1485, in fit_generator callbacks.on_epoch_end(epoch, epoch_logs) File "/home/ashok/.local/lib/python2.7/site-packages/keras/callbacks.py", line 40, in on_epoch_end callback.on_epoch_end(epoch, logs) File "/home/ashok/.local/lib/python2.7/site-packages/keras/callbacks.py", line 296, in on_epoch_end self.model.save(filepath, overwrite=True) File "/home/ashok/.local/lib/python2.7/site-packages/keras/engine/topology.py", line 2423, in save save_model(self, filepath, overwrite) File "/home/ashok/.local/lib/python2.7/site-packages/keras/models.py", line 48, in save_model f = h5py.File(filepath, 'w') File "/usr/local/lib/python2.7/dist-packages/h5py/_hl/files.py", line 271, in init fid = make_fid(name, mode, userblock_size, fapl, swmr=swmr) File "/usr/local/lib/python2.7/dist-packages/h5py/_hl/files.py", line 107, in make_fid fid = h5f.create(name, h5f.ACC_TRUNC, fapl=fapl, fcpl=fcpl) File "h5py/_objects.pyx", line 54, in h5py._objects.with_phil.wrapper (/tmp/pip-nCYoKW-build/h5py/_objects.c:2840) File "h5py/_objects.pyx", line 55, in h5py._objects.with_phil.wrapper (/tmp/pip-nCYoKW-build/h5py/_objects.c:2798) File "h5py/h5f.pyx", line 98, in h5py.h5f.create (/tmp/pip-nCYoKW-build/h5py/h5f.c:2284) IOError: Unable to create file (Unable to open file: name = '/home/ashok/desktop/out/modelweights/cnnmodeldep80weights.best.hdf5', errno = 2, error message = 'no such file or directory', flags = 13, o_flags = 242)

Process finished with exit code 1

hi, the error which is posted was not resolving .so,could you give me the suggestion to solve it.

Thank you.

I can't understand why the model is not able to save in the given path .

I am on traveling. I will check the code for you later

What version of keras you are using?

If you are using tensorflow, make sure to make some modifications to the original code to be compatible with it

i am using keras 1.1.1 and theano 0.8.2 versions .I have modified the in some lines and comment the code of Prepare data sets as we i have the complete data and the train and test data also. the code i have do some modifications in your code is below.

-- coding: utf-8 --

""" Classification on Small-scale Image Data based on Data Expansion/Augmentation Created on Wed Nov 09 22:25:36 2016

Edit history: 11/20/2016: add early stopping and modelcheckpoint
11/22/2016: data splitting of 80% training and 20% testing Three training cases: 1) 100 epochs 2) 200 epochs 3) 400 epochs 4) 600 epochs Obtained best accuracies 1): 83.37% for training and 87.24% for testing 2): 87.24% for training and 89.21% for testing 3): 90.34% for training and 90.41% for testing 4): 93.08% for training and 89.98% for testing

@author: Yuzhen Lu """

from keras.preprocessing.image import ImageDataGenerator, img_to_array, load_img import matplotlib.pyplot as plt

'''datagen = ImageDataGenerator( rotation_range=40, width_shift_range=0.2, height_shift_range=0.2, zoom_range=0.2, horizontal_flip=True, fill_mode='nearest')

from PIL import Image

img = Image.open('Test/apple/4.jpeg')

img = load_img('imagePath') # this is a PIL image img.save('saveImagePath') plt.imshow(img) x = img_to_array(img) # this is a Numpy array with shape (3, 128, 128) x = x.reshape((1,) + x.shape)

###############################################################################

Prepare data sets

###############################################################################

Split images into training and testing parts

Note: run only one time!!

def splitImageSet(rootFolder, outFolder, p): import os from os import listdir import numpy as np from PIL import Image cats = listdir(rootFolder) # a list of subfolder names
for cat in cats: print('Image Category...{}'.format(cat))

    folderPath = (os.path.join(rootFolder,cat))
    imgNames = listdir(folderPath)
    imgPaths = [os.path.join(folderPath,imgName) for imgName in imgNames]
    idx = np.random.permutation(len(imgPaths))
    trainIdx = idx[:int(p*len(idx))]
    testIdx = [ind for ind in idx if not ind in trainIdx]

    if not os.path.exists(os.path.join(outFolder,'Train',cat)):
        os.makedirs(os.path.join(outFolder,'Train',cat))
    for k in range(len(trainIdx)):
        img = Image.open(os.path.join(imgPaths[trainIdx[k]]))
        #temp = os.path.join(outFolder,'train',cat,imgNames[trainIdx[k]])
        img.save(os.path.join(outFolder,'Train',cat,imgNames[trainIdx[k]]))            
    if not os.path.exists(os.path.join(outFolder,'Test',cat)):
        os.makedirs(os.path.join(outFolder,'Test',cat))
    for k in range(len(testIdx)):
        img = Image.open(os.path.join(imgPaths[testIdx[k]]))
        img.save(os.path.join(outFolder,'Test',cat,imgNames[testIdx[k]])) 

print('Split Done!')
return

rootFolder = 'rootFolder' #add the image directory outFolder = 'outFolder' #image output directory
splitImageSet(rootFolder, outFolder, 0.80)

Please start from here!!'''

###############################################################################

Build a CNN model

###############################################################################

CNN model

import os import numpy as np from keras.models import Sequential from keras.layers import Convolution2D, MaxPooling2D from keras.layers import Activation, Dropout, Flatten, Dense from keras.optimizers import SGD from keras.callbacks import EarlyStopping, ModelCheckpoint, History from keras import backend as K

model = Sequential() model.add(Convolution2D(32, 7, 7, input_shape=(128, 128, 3))) model.add(Activation('relu')) model.add(MaxPooling2D(pool_size=(2, 2))) model.add(Convolution2D(64, 5, 5)) model.add(Activation('relu')) model.add(MaxPooling2D(pool_size=(2, 2))) model.add(Convolution2D(128, 3, 3)) model.add(Activation('relu')) model.add(MaxPooling2D(pool_size=(2, 2))) model.add(Dropout(0.25)) model.add(Flatten()) # this converts our 3D feature maps to 1D feature vectors model.add(Dense(128)) model.add(Activation('relu')) model.add(Dropout(0.5)) model.add(Dense(10)) model.add(Activation('softmax')) sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True) #initial lr = 0.01 model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy']) print(model.summary())

####################################

Callback Schedule

################################### import keras class decaylr_loss(keras.callbacks.Callback): def init(self): super(decaylr_loss, self).init() def on_epoch_end(self,epoch,logs={}):

loss=logs.items()[1][1] #get loss

    loss=logs.get('loss')
    print("loss: ",loss)
    old_lr = 0.001 #needs some adjustments
    new_lr= old_lr*np.exp(loss) #lr*exp(loss)
    print("New learning rate: ", new_lr)
    K.set_value(self.model.optimizer.lr, new_lr)

lrate = decaylr_loss()

early stopping

patience = 20 earlystopper = EarlyStopping(monitor='val_acc', patience=patience, verbose=1, mode='max')

check point

wdir = '/home/ashok/Desktop/out' #work directory filepath = os.path.join(wdir,'modelWeights','cnnModelDEp80weights.best.hdf5') #save model weights print('hello',filepath) checkpoint = ModelCheckpoint(filepath, monitor='val_acc', verbose=1, save_best_only=True, mode='max')

###############################################################################

Data Expansion or Augmentation

############################################################################### from keras.preprocessing.image import ImageDataGenerator train_datagen = ImageDataGenerator(

featurewise_center=True,

                               #featurewise_std_normalization=True,
                               rescale=1./255,
                               rotation_range=20,
                               width_shift_range=0.2,
                               height_shift_range=0.2,
                               zoom_range=0.2,
                               horizontal_flip=True)

trainDir = './Train' train_generator = train_datagen.flow_from_directory(trainDir, target_size=(128,128), batch_size=32, class_mode='categorical') test_datagen = ImageDataGenerator(rescale=1./255) testDir = './Test' test_generator = test_datagen.flow_from_directory(testDir, target_size=(128,128), batch_size=32, shuffle=False, class_mode='categorical')

###############################################################################

Fit, Evaluate and Save Model

############################################################################### epochs = 1

epochs = 200

epochs = 400

epochs = 600

samples_per_epoch = 4654 val_samples = 1168

Fit the model

hist = History() model.fit_generator(train_generator, samples_per_epoch= samples_per_epoch, nb_epoch=epochs, verbose=1, validation_data=test_generator, nb_val_samples=val_samples, callbacks = [earlystopper, lrate, checkpoint, hist])

evaluate the model

scores = model.evaluate_generator(test_generator, val_samples=val_samples) print("Accuracy = ", scores[1])

save model

savePath = wdir print('hhh',savePath) model.save_weights(os.path.join(savePath,'cnnModelDEp80.h5')) # save weights after training or during training model.save(os.path.join(savePath,'cnnModelDEp80.h5')) #save complied model

plot acc and loss vs epochs

import matplotlib.pyplot as plt print(hist.history.keys())

accuracy

plt.plot(hist.history['acc']) plt.plot(hist.history['val_acc']) plt.title('model accuracy') plt.ylabel('accuracy') plt.xlabel('epoch') plt.legend(['train', 'test'], loc='best') plt.savefig(os.path.join(savePath,'cmdeP80AccVsEpoch.jpeg'), dpi=1000, bbox_inches='tight') plt.show()

loss

plt.plot(hist.history['loss']) plt.plot(hist.history['val_loss']) plt.title('model loss') plt.ylabel('loss') plt.xlabel('epoch') plt.legend(['train', 'test'], loc='best') plt.savefig(os.path.join(savePath,'cmdeP80LossVsEpoch.jpeg'), dpi=1000, bbox_inches='tight') plt.show()

###############################################################################

Note: train 4364 images (80%) and test 1458 images (20%)

100 epochs:

200 epoches: acc = 0.8724; val_acc = 0.89212

400 epoches:

600 epochs:

###############################################################################

load the model

not necessary the best at the end of training

from keras.models import load_model myModel = load_model(os.path.join(savePath,'cnnModelDEp80.h5'))
scores = myModel.evaluate_generator(test_generator,val_samples) print("Accuracy = ", scores[1])

########################

Check-pointed model

####################### model = Sequential() model.add(Convolution2D(32, 7, 7, input_shape=(3, 128, 128))) model.add(Activation('relu')) model.add(MaxPooling2D(pool_size=(2, 2))) model.add(Convolution2D(64, 5, 5)) model.add(Activation('relu')) model.add(MaxPooling2D(pool_size=(2, 2))) model.add(Convolution2D(128, 3, 3)) model.add(Activation('relu')) model.add(MaxPooling2D(pool_size=(2, 2))) model.add(Dropout(0.25)) model.add(Flatten()) model.add(Dense(128)) model.add(Activation('relu')) model.add(Dropout(0.5)) model.add(Dense(10)) model.add(Activation('softmax'))

lr = 0.00277615583366 #adjust lr based on training process

lr = 0.00181503843077

lr = 0.00163685841542 #case 2

lr = 0.00122869861281 # case3 sgd = SGD(lr=lr, decay=1e-6, momentum=0.9, nesterov=True) model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy']) model.load_weights(filepath) #load saved weights scores = model.evaluate_generator(test_generator,val_samples) print("Accuracy = ", scores[1])

Confusion matrix on the test images

imgDir = testDir

imgDir = trainDir test_generator = test_datagen.flow_from_directory(imgDir, target_size=(128,128), batch_size=32, shuffle=False, class_mode='categorical')

val_samples = 1168

val_samples = 4654 predict = model.predict_generator(test_generator,val_samples)

yTrue = test_generator.classes yTrueIdx = test_generator.class_indices

from sklearn.metrics import classification_report, confusion_matrix yHat = np.ones(predict.shape[0],dtype = int) for i in range(predict.shape[0]): temp = predict[i,:] yHat[i] = np.argmax(temp)

from sklearn.metrics import accuracy_score acc = accuracy_score(yTrue,yHat) print("Accuracy on test images:", acc) #same as scores[1]

def numToLabels(y,cat): numLabel = [] import numpy as np yNew = np.unique(y) #sorted for i in range(len(y)): idx = np.where(yNew == y[i])[0][0] numLabel.append(cat[idx]) return numLabel

labels = sorted(yTrueIdx.keys())

labels = ['Ap','Ba','Br','Bu','Eg','Fr','Hd','Pz','Rc','St'] yActLabels = numToLabels(yTrue,labels) yHatLabels = numToLabels(yHat,labels) CM = confusion_matrix(yActLabels,yHatLabels,labels) #np.array

print CM

print(classification_report(yTrue,yHat,target_names=labels))

Alternatively: pd.crosstab

import pandas as pd

preds = pd.DataFrame(predict)

y1 = pd.Categorical(yActLabels,categories=labels) y2 = pd.Categorical(yHatLabels,categories=labels) pd.crosstab(y1,y2,rownames=['True'], colnames=['Predicted'])

###############################################################################

Miscellaneous

###############################################################################

evaluate execution efficiency

import time t = time.time() s = model.predict_generator(test_generator,val_samples) elapsedTime = time.time()-t print('Average time: {} second'.format(elapsedTime/val_samples))

average time: 0.00895 s, less than 0.01 s

def getImgPaths2(rootPath): import os from os import listdir print('Extract paths and labels...') cats = listdir(rootPath) imgPaths = [] imgLabels = [] for cat in cats: print('{}...'.format(cat)) foldPaths = os.path.join(rootPath, cat) imgPaths.extend([os.path.join(foldPaths,imgName) for imgName in listdir(foldPaths)]) imgLabels.extend([cat]len(listdir(foldPaths))) return imgPaths, imgLabels def getImgData(imgPaths): from scipy import misc import numpy as np print('Extract image data...') temp1 = misc.imread(imgPaths[0]) imgData = np.zeros((len(imgPaths),temp1.shape[0],temp1.shape[1],temp1.shape[2]), dtype='float32') for ii in range(len(imgPaths)): temp = misc.imread(imgPaths[ii]) imgData[ii,:,:,:] = temp print("\r{} Percent complete\r".format(100(ii+1)/len(imgPaths)),) return imgData imgPaths = getImgPaths2(trainDir)

Expanded image data

for i in range(0, 9): plt.subplot(330 + 1 + i) plt.imshow() plt.imshow()

#######################################################

Images with correct or wroning predicted Labels

####################################################### imgPaths, imgLabels = getImgPaths2(testDir) X_test = getImgData(imgPaths) #(1168, 128, 128, 3) X_test /= 255 test_wrong = [im for im in zip(X_test, yHatLabels, yActLabels) if im[1] != im[2]] print(len(test_wrong))

112 misclassified images

show some misclassified images

import matplotlib.pyplot as plt plt.figure(figsize=(7, 8)) import numpy as np for ind, val in enumerate(test_wrong):

print ind

plt.subplots_adjust(left=0, right=1, top=1, bottom=0)
plt.subplot(7, 8, ind + 1)
img = val[0]
img *= 255
plt.axis("off")
plt.text(0,0,val[2], fontsize=12, color='blue')
plt.text(40,0,val[1], fontsize=12, color='red')
plt.imshow(img.astype('uint8'))
if ind==55:
    break

plt.savefig(os.path.join(savePath,'MissClassifiedImages1.jpeg'), dpi=800, bbox_inches='tight') plt.show()

show some misclassified images

import matplotlib.pyplot as plt plt.figure(figsize=(7, 8)) import numpy as np for ind, val in enumerate(test_wrong[56:]):

print ind

plt.subplots_adjust(left=0, right=1, top=1, bottom=0)
plt.subplot(7, 8, ind + 1)
img = val[0]
img *= 255
plt.axis("off")
plt.text(0,0,val[2], fontsize=12, color='blue')
plt.text(40,0,val[1], fontsize=12, color='red')
plt.imshow(img.astype('uint8'))
if ind==55:
    break

plt.savefig(os.path.join(savePath,'MissClassifiedImages2.jpeg'), dpi=800, bbox_inches='tight') plt.show()

########################################

Rotated, shifted, sheared Images

######################################## imgData = getImgData(imgPaths[:9]) #extract 9 images, (9L, 128L, 128L, 3L)

plot raw images

for i in range(0, 9): from scipy import misc img = misc.imread(imgPaths[i]) plt.subplot(3,3,i+1) fig = plt.imshow(img) ax = plt.gca()

ax.set_axis_off()

ax.axes.get_xaxis().set_ticks([])
ax.axes.get_yaxis().set_ticks([])

plt.show()

or

for i in range(0, 9): img = imgData[i,:,:,:].astype('uint8') plt.subplot(3,3,i+1) fig = plt.imshow(img) ax = plt.gca() ax.set_axis_off() plt.savefig(os.path.join(savePath,'RawImages.jpeg'), dpi=1000, bbox_inches='tight') plt.show()

image distortion by ImageDataGenerator

from keras.preprocessing.image import ImageDataGenerator #, array_to_img, img_to_array datagen = ImageDataGenerator( rotation_range=20, width_shift_range=0.2, height_shift_range=0.2, zoom_range=0.2, horizontal_flip=True )

single fruit

from PIL import Image

x = imgData[1].reshape((1,)+imgData[0].shape) fig = plt.imshow(imgData[1].astype('uint8')) #imshow: img_dim_ordering (w,h,channels) plt.gca().set_axis_off() plt.savefig(os.path.join(savePath,'rawImage.jpeg'), dpi=1000, bbox_inches='tight') x = x.transpose(0,3,1,2) #reshaped into 4d for datagen.flow i = 0 for x_batch in datagen.flow(x, batch_size=1): i += 1 plt.subplot(3,3,i) I = x_batch[0] img = (I.transpose(1,2,0)).astype('uint8') fig=plt.imshow(img,cmap=plt.get_cmap('gray')) plt.gca().set_axis_off() if i==9: break plt.savefig(os.path.join(savePath,'expandedImages1.jpeg'), dpi=1000, bbox_inches='tight')

multiple fruit

for x_batch in datagen.flow(imgData.transpose(0,3,1,2), batch_size=9): for i in range(0,9): plt.subplot(3,3,i+1) I = x_batch[i]

img = array_to_img(I)

    img = (I.transpose(1,2,0)).astype('uint8')
    fig = plt.imshow(img,cmap=plt.get_cmap('gray'))
    plt.gca().set_axis_off()
plt.show()
break

plt.savefig(os.path.join(savePath,'expandedImages2.jpeg'), dpi=1000, bbox_inches='tight')

evaluate execution efficiency

import time t = time.time() s = model.predict_generator(test_generator,val_samples) elapsedTime = time.time()-t print ('Average time: {} second'.format(elapsedTime/val_samples))

average time: 0.00895 s, less than 0.01 s

Hi ,can you guide me.

can you please do the favour for me,as I am not getting the model.

How are you , How you doing?

As we discussed about the code of food classification error (i.e about the model not saving in the given path or default path).

Please check the code for me.

Thank you , Vikash

On Thu, Aug 10, 2017 at 5:48 PM, jingweimo notifications@github.com wrote:

I am on traveling. I will check the code for you later

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/jingweimo/food-image-classification-/issues/1#issuecomment-321535039, or mute the thread https://github.com/notifications/unsubscribe-auth/AdLFOty7GO7nMVYQM-4SBQAtl9bk9QH5ks5sWvUvgaJpZM4OoErJ .

jingweimo / food-image-classification-

Error occurring while running the code #1

Layer (type) Output Shape Param #

activation_5 (Activation) (None, 10) 0

Layer (type) Output Shape Param # Connected to

activation_5 (Activation) (None, 10) 0 dense_2[0][0]

-- coding: utf-8 --

from PIL import Image

img = Image.open('Test/apple/4.jpeg')

Prepare data sets

Split images into training and testing parts

Note: run only one time!!

Please start from here!!'''

Build a CNN model

CNN model

Callback Schedule

loss=logs.items()[1][1] #get loss

early stopping

check point

Data Expansion or Augmentation

featurewise_center=True,

Fit, Evaluate and Save Model

epochs = 200

epochs = 400

epochs = 600

Fit the model

evaluate the model

save model

plot acc and loss vs epochs

accuracy

loss

Note: train 4364 images (80%) and test 1458 images (20%)

100 epochs:

200 epoches: acc = 0.8724; val_acc = 0.89212

400 epoches:

600 epochs:

load the model

not necessary the best at the end of training

Check-pointed model

lr = 0.00277615583366 #adjust lr based on training process

lr = 0.00181503843077

lr = 0.00163685841542 #case 2

Confusion matrix on the test images

imgDir = testDir

val_samples = 1168

labels = sorted(yTrueIdx.keys())

print CM

Alternatively: pd.crosstab

preds = pd.DataFrame(predict)

Miscellaneous

evaluate execution efficiency

average time: 0.00895 s, less than 0.01 s

Expanded image data

Images with correct or wroning predicted Labels

112 misclassified images

show some misclassified images

print ind

show some misclassified images

print ind

Rotated, shifted, sheared Images

plot raw images

ax.set_axis_off()

or

image distortion by ImageDataGenerator

single fruit

from PIL import Image

multiple fruit

img = array_to_img(I)

evaluate execution efficiency

average time: 0.00895 s, less than 0.01 s