georgeseif / transfer-learning-suite Goto Github PK

Hi，GeorgeSeif：
Thanks for your transfer learning code，if I want to setting different learning rate in different layer?Is keras
possible to do this ?
Thanks!

When I make my own dataset , and try to run main.py , there is an error.

I created structure as :
-'dataset'
-train
-class1
(include images)
-class2
-val
-class1
-class2

Details of the error followed:
Found 705 images belonging to 2 classes.
Found 361 images belonging to 2 classes.
Epoch 1/100
Traceback (most recent call last):
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/site-packages/keras/utils/data_utils.py", line 578, in get
inputs = self.queue.get(block=True).get()
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/multiprocessing/pool.py", line 644, in get
raise self._value
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/multiprocessing/pool.py", line 119, in worker
result = (True, func(*args, **kwds))
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/site-packages/keras/utils/data_utils.py", line 401, in get_index
return _SHARED_SEQUENCES[uid][i]
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/site-packages/keras/preprocessing/image.py", line 825, in getitem
return self._get_batches_of_transformed_samples(index_array)
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/site-packages/keras/preprocessing/image.py", line 1233, in _get_batches_of_transformed_samples
img = self.image_data_generator.preprocessing_function(img)
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/site-packages/keras/applications/imagenet_utils.py", line 178, in preprocess_input
mode=mode)
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/site-packages/keras/applications/imagenet_utils.py", line 132, in _preprocess_symbolic_input
x = x[..., ::-1]
TypeError: 'Image' object is not subscriptable

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "/home/he/pycharm/pycharm-2017.3.4/helpers/pydev/pydevd.py", line 1668, in
main()
File "/home/he/pycharm/pycharm-2017.3.4/helpers/pydev/pydevd.py", line 1662, in main
globals = debugger.run(setup['file'], None, None, is_module)
File "/home/he/pycharm/pycharm-2017.3.4/helpers/pydev/pydevd.py", line 1072, in run
pydev_imports.execfile(file, globals, locals) # execute the script
File "/home/he/pycharm/pycharm-2017.3.4/helpers/pydev/_pydev_imps/_pydev_execfile.py", line 18, in execfile
exec(compile(contents+"\n", file, 'exec'), glob, loc)
File "/home/he/Code/Transfer-Learning-Suite/main.py", line 206, in
validation_data=validation_generator, validation_steps=num_val_images // BATCH_SIZE, class_weight='auto', shuffle=True, callbacks=callbacks_list)
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/site-packages/keras/legacy/interfaces.py", line 91, in wrapper
return func(*args, **kwargs)
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/site-packages/keras/engine/training.py", line 2192, in fit_generator
generator_output = next(output_generator)
File "/home/he/anaconda2/envs/Yolo-keras/lib/python3.5/site-packages/keras/utils/data_utils.py", line 584, in get
six.raise_from(StopIteration(e), e)
File "", line 3, in raise_from
StopIteration: 'Image' object is not subscriptable

Thank you so much.

how to run prediction on test data folder?

OS Platform and Distribution: windows 10
TensorFlow backend: yes
tensorflow 1.13.1
tensorflow-base 1.13.1
tensorflow-estimator 1.13.0
Keras version:
keras 2.2.4
keras-applications 1.0.7
keras-base 2.2.4
keras-preprocessing 1.0.9
Python version:
Python 3.6.5
No cuda
No gpu

On running the code as described, I am getting-
Traceback (most recent call last): File "main.py", line 205, in <module> validation_data=validation_generator, validation_steps=1, class_weight='auto', shuffle=True, callbacks=callbacks_list) File "C:\Users\Win 10\AppData\Local\Programs\Python\Python36\lib\site-packages\keras\legacy\interfaces.py", line 91, in wrapper return func(*args, **kwargs) File "C:\Users\Win 10\AppData\Local\Programs\Python\Python36\lib\site-packages\keras\engine\training.py", line 1418, in fit_generator initial_epoch=initial_epoch) File "C:\Users\Win 10\AppData\Local\Programs\Python\Python36\lib\site-packages\keras\engine\training_generator.py", line 251, in fit_generator callbacks.on_epoch_end(epoch, epoch_logs) File "C:\Users\Win 10\AppData\Local\Programs\Python\Python36\lib\site-packages\keras\callbacks.py", line 79, in on_epoch_end callback.on_epoch_end(epoch, logs) File "C:\Users\Win 10\AppData\Local\Programs\Python\Python36\lib\site-packages\keras\callbacks.py", line 338, in on_epoch_end self.progbar.update(self.seen, self.log_values) AttributeError: 'ProgbarLogger' object has no attribute 'log_values'

@GeorgeSeif
This is an interesting project. I tried this with dog vs. cat dataset (kaggle) that was running fine. The only problem was the error in plotting.

Epoch 00019: saving model to ./checkpoints/DenseNet201_model_weights.h5
Epoch 20/20
703/703 [==============================] - 149s 212ms/step - loss: 0.0014 - acc: 0.9996 - val_loss: 0.1408 - val_acc: 0.9764

Epoch 00020: saving model to ./checkpoints/DenseNet201_model_weights.h5
Traceback (most recent call last):
File "main.py", line 208, in
plot_training(history)
NameError: name 'plot_training' is not defined

Hi, I am trying to make a predictions, but looks like the function call is not complete.
No significant changes in code was done. I am trying it with python3

Was trained with:
python main.py --mode train --dataset new_train_3 --model InceptionResNetV2

I am trying to predict with:
python main.py --mode predict --image ~/Devel/ai-dev/IMGS/999f2cfbfbde894166bb9a4bbf8df46d.640x360.jpg --model InceptionResNetV2 --dataset new_train_3

How can I get it work?

Nice Work!

Cleanest solution that exists so far.

Any plans to introduce object detection/ bounding box for e.g YOLO?

Hey! Great code by the way, but I want to point out an error. If I try to use InceptionResnetV2, I get "ModuleNotFoundError: No module named 'keras.applications.inceptionresnetv2'"

To fix that change

keras.applications.inceptionresnetv2

to

keras.application.inception_resnet_v2

in your main.py so that it would look like that:

elif args.model == "InceptionResNetV2":
    from keras.applications.inception_resnet_v2 import preprocess_input
    preprocessing_function = preprocess_input
    base_model = InceptionResNetV2(weights='imagenet', include_top=False, input_shape=(HEIGHT, WIDTH, 3))

Should probably be there in the readme?

Hi George,
after preparing my own dataset i run into the following error.

https://i.imgur.com/1WUFlHa.png

Noting the numbers are very close, i tried different models and batch sizes but can not make progress. Do you have an idea what might cause the error?

This is the relevant line from main.py
history = finetune_model.fit_generator(train_generator, epochs=args.num_epochs, workers=8, steps_per_epoch=num_train_images // BATCH_SIZE, validation_data=validation_generator, validation_steps=num_val_images // BATCH_SIZE, class_weight='auto', shuffle=True, callbacks=callbacks_list)