search

Visual-Attention-Network / SegNeXt

Official Pytorch implementations for "SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation" (NeurIPS 2022)
Apache License 2.0
789 stars 86 forks source link

The content of the article is inconsistent with the code #46

Open Flyingdog-Huang opened 1 year ago

Flyingdog-Huang commented 1 year ago

In codes, inputs = torch.cat(inputs, dim=1) https://github.com/Visual-Attention-Network/SegNeXt/blob/c87bcaef22ed2ad72ed3decb3727e5f879343516/mmseg/models/decode_heads/ham_head.py#L233

In paper, our decoder only receives features from the last three stages

can you tell me what happened

3maoyap commented 1 year ago

You should learn the parameter "in_index".