TypeError: 'numpy.float64' object cannot be interpreted as an index

东方耀

fg_inds = npr.choice(fg_inds, size=fg_rois_per_this_image, replace=False)
  File "mtrand.pyx", line 1197, in mtrand.RandomState.choice
TypeError: 'numpy.float64' object cannot be interpreted as an index




py-faster-rcnn/lib/roi_data_layer/minibatch.py  修改第26行 为：
fg_rois_per_image = np.round(cfg.TRAIN.FG_FRACTION * rois_per_image).astype(np.int)




py-faster-rcnn/lib/datasets/ds_utils.py  修改第13行  为：
hashes = np.round(boxes * scale).astype(np.int).dot(v)




py-faster-rcnn/lib/fast_rcnn/test.py  修改第134行  为：
hashes = np.round(blobs['rois'] * cfg.DEDUP_BOXES).astype(np.int).dot(v)




py-faster-rcnn/lib/rpn/proposal_target_layer.py 修改第60行  为：
fg_rois_per_image = np.round(cfg.TRAIN.FG_FRACTION * rois_per_image).astype(np.int)