pytorch / pytorch/vision

Enable Video models for other tasks

Open
#8,400 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

🚀 The feature

It would be nice to enable video models to be used as encoders for other tasks then classification.

Motivation, pitch

It seems that video models (e.g. mvit, swin video..) are strictly connected internally to their classification head.
It could be useful to let use them in others tasks. Just to make an example we could use them as a multi-scale encoder to be connected to other components like an encoder-decoder architecture for video segmentation, 3d medical segmentation etc..

Alternatives

No response

Additional context

No response

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

The issue points to torchvision/models/video; start by reviewing the video model implementations and how they are coupled to classification heads. Determine the interface needed for using these models as encoders, and consider how support would serve the proposed segmentation use cases. Done means video models can be used beyond classification.

Written by the indexing model from the issue text.

Assessment

Tech stack
python, pytorch
Domain
computer-vision, machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
30/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.