Training fails after some epochs passed. about keras-fasterrcnn HOT 2 OPEN

@saleed unfortunately, I do not have a solution to this. I just wrapped the whole training block with try except in order to continue training, as a result when the exception occurs it skips the training step. This error might be related to these lines in RoiPoolingConv :

for roi_idx in range(self.num_rois):
    x = rois[0, roi_idx, 0]
    y = rois[0, roi_idx, 1]
    w = rois[0, roi_idx, 2]
    h = rois[0, roi_idx, 3]

If you inspect this issue #42 it has the full traceback and I think it points to these lines. I thought, writing a for each loop in these lines would solve the problem but never got to test it out. Indeed roi_idx might be out of bounds here and I do not know if we are guaranteed to have self.num_rois number of ROIs at each training step.

from keras-fasterrcnn.

I got the same error, have you solved this problem?

from keras-fasterrcnn.