mlearing / keras-faster-rcnn Goto Github PK

I need to decrease the processing time and used RAM
so I thought that I can deal with gray scale images by reading the image as gray scale image using imread parameters
img = cv2.imread(filename,0)
image_parser.py
data_augment.py

so the dimensions would be
input_shape_img = (None, None, 1)
instead of
input_shape_img = (None, None, 3)
train.py

but I field in optimizing the model
I retrieve an exception on this code
loss_rpn = model_rpn.train_on_batch(X, Y)
train.py