Hello Jeasinema,
I go through the codes these days and find a issue during calculate the ih in bbox_overlaps function, your code are:
ih = (
min(boxes[n, 3], query_boxes[k, 3]) -
max(boxes[n, 1], query_boxes[k, 1]) + 1
)
but I think it should be:
ih = (
min(boxes[n, 1], query_boxes[k, 1]) -
max(boxes[n, 3], query_boxes[k, 3]) + 1
)
wish your feedback, thanks.
quick link:
https://github.com/tsinghua-rll/VoxelNet-tensorflow/blob/d24c3eefc5762e4785ca3be6ca7d1fd0ef16bc61/utils/box_overlaps.pyx#L46
Hello Jeasinema, I go through the codes these days and find a issue during calculate the ih in bbox_overlaps function, your code are: ih = ( min(boxes[n, 3], query_boxes[k, 3]) - max(boxes[n, 1], query_boxes[k, 1]) + 1 ) but I think it should be: ih = ( min(boxes[n, 1], query_boxes[k, 1]) - max(boxes[n, 3], query_boxes[k, 3]) + 1 ) wish your feedback, thanks. quick link: https://github.com/tsinghua-rll/VoxelNet-tensorflow/blob/d24c3eefc5762e4785ca3be6ca7d1fd0ef16bc61/utils/box_overlaps.pyx#L46