dongluyang

could you explain much details about the logic for fpn instance segmentation as below. what is MLP, and what does figure 4 mean? Figure 4. FPN for object segment proposals....

FYI, below links give my study for LaTex-OCR, hope it will be useful for your use https://www.toutiao.com/article/7259700493524402707/

C:\Users\panpa\ai\pytorch_simple_CornerNet\utils\losses.py:61: UserWarning: indexing with dtype torch.uint8 is now deprecated, please use a dtype torch.bool instead. (Triggered internally at C:\cb\pytorch_1000000000000\work\aten\src\ATen/native/IndexingUtils.h:28.) loss = sum([F.smooth_l1_loss(r[mask], gt_regs[mask], reduction='sum') / num for r in regs])...

File "C:\Users\panpa\anaconda3\lib\site-packages\torch\nn\functional.py", line 2438, in batch_norm return torch.batch_norm( RuntimeError: CUDA out of memory. Tried to allocate 768.00 MiB (GPU 0; 24.00 GiB total capacity; 22.52 GiB already allocated; 0 bytes...