Squeeze and excitation (scSE) modules
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 2.1k
- Forks
- 385
- PR merge metrics
- No merged PRs in 30d
Description
We should look into squeeze and excitation (SE) modules for adaptive feature map recalibration.
The initially introduces cSE blocks are incredibly powerful for classification problems; the recently introduces sSE blocks are similarly powerful for segmentation.
We should try a combination of these in an scSE module and see what happens if we attach them to our encoder and decoder modules. From a computational standpoint they are very cheap.
References below
https://arxiv.org/abs/1709.01507
> Squeeze-and-Excitation Networks
https://arxiv.org/abs/1803.02579
> Concurrent Spatial and Channel 'Squeeze & Excitation' in Fully Convolutional Networks
_from https://arxiv.org/abs/1803.02579_
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
The issue names no files, tests, or entry point. Start by reading the two referenced SE papers and locating the encoder and decoder modules; done would require an agreed scSE design, integration with those modules, and classification or segmentation experiment results.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- computer-vision, machine-learning
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100