pytorch / pytorch/vision

Can ResNet18 / BasicBlock support replace_stride_with_dilation = [False, False, True]?

Open
#4,801 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
17.9k
Forks
7.3k
Avg merge
1d 15h
Merged PRs (30d)
13

Description

🐛 Describe the bug

Currently, it doesn't work because of https://github.com/pytorch/vision/blob/main/torchvision/models/resnet.py#L77

But I guess, one could just dilate conv1? (just like conv2 is dilated in Bottleneck)

e.g. this is how it's done in ResNet-WS: https://github.com/shenyunhang/UWSOD/blob/UWSOD/projects/WSL/wsl/modeling/backbone/resnet_ws.py#L32

Versions

N/A

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start in torchvision/models/resnet.py at line 77 and trace how ResNet18 and BasicBlock handle replace_stride_with_dilation. Compare the proposed conv1 dilation behavior with the linked ResNet-WS implementation. Done means the [False, False, True] configuration works without the current failure and preserves existing configurations.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
computer-vision, machine-learning
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.