Yujia-Tang

Use own data to train this model

1

I try to train my own data using this model, but there is easy to be NaN loss. I have modified my learning rate enough small and double the batch...

When calculate IOU between bbox and other groudtruth. We always use iou(bbox[1],groundtruth)(here bbox and groundtruth are all [n,4]). For batch-size=1, no probelm here. However, for batch-size>1, we cannot use this...

When calculate IOU between bbox and other groudtruth. We always use iou(bbox[1],groundtruth)(here bbox and groundtruth are all [n,4]). For batch-size=1, no probelm here. However, for batch-size>1, we cannot use this...