weecology / weecology/DeepForest

Explore bioclip for crop model backbone

Open
#899 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
774
Forks
265
PR merge metrics
No merged PRs in 30d

Description

As part of the larger effort to integrate among different research programs, I think we could use bioclip as the basis for fine-tuning for the cropmodel

https://github.com/Imageomics/bioclip/tree/main

Questions

  1. Would we need to wrap this repo, or a static version inside deepforest.
  2. Add open clip dependency, will this be an issue?
  3. Should be add this as a submodule installation. pip install DeepForest[bioclip]
  4. How does finetuning this versus a simple resnet perform for airborne imagery?

I started down this road in a toy repo here: https://github.com/weecology/BioClip_FieldGuide/blob/5789a7fc023f7cc755cd1700b7a4b90d6e192f83/bioclip_example/FieldGuide.py#L18

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 with the linked bioclip repository and the toy example in bioclip_example/FieldGuide.py at the referenced revision. Clarify whether DeepForest should wrap bioclip, vendor it, or expose an optional installation, then define a reproducible comparison between bioclip fine-tuning and a ResNet on airborne imagery. Done requires an agreed integration path, dependency decision, and evaluation plan.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.