Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Scouting vision feature extractor #10

Closed
3 tasks
Francesco-Ranieri opened this issue Mar 13, 2023 · 3 comments
Closed
3 tasks

Scouting vision feature extractor #10

Francesco-Ranieri opened this issue Mar 13, 2023 · 3 comments
Assignees
Labels
enhancement New feature or request

Comments

@Francesco-Ranieri
Copy link
Collaborator

Francesco-Ranieri commented Mar 13, 2023

Focal points:

  • add features extraction to DVC pipeline
  • add mlflow library to manage the ML lifecycle and track experiments
  • try different pre trained model for vision features extraction:
    • DETR : facebook/deit-base-patch16-224
    • CLIP: openai/clip-vit-large-patch14-336
    • VITC: google/vit-large-patch16-224-in21k
@Francesco-Ranieri Francesco-Ranieri added the enhancement New feature or request label Mar 13, 2023
@Francesco-Ranieri Francesco-Ranieri self-assigned this Mar 13, 2023
@dagshub
Copy link

dagshub bot commented Mar 13, 2023

@Francesco-Ranieri
Copy link
Collaborator Author

The code used for this scouting is this one amazon/mm-cot/issue#35.

@gianfrancodemarco gianfrancodemarco changed the title Scounting vision feature extractor Scouting vision feature extractor Mar 13, 2023
@gianfrancodemarco
Copy link
Owner

gianfrancodemarco commented Mar 13, 2023

@Francesco-Ranieri we could implement a simple DVC piepeline with a step for the extraction of vision features. We can 2 parameters to chose the model to use for the extraction and the path to the dataframe. What do you think?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

When branches are created from issues, their pull requests are automatically linked.

2 participants