README.md

metadata

title: MCA Classifier
emoji: 🐠
colorFrom: indigo
colorTo: purple
sdk: gradio
sdk_version: 5.37.0
app_file: app.py
pinned: false
license: cc-by-4.0
short_description: Multi-Crop Attention-based Classifier

Paying Attention to Other Animal Detections Improves Camera Trap Classification Models

LBBE; CEFE; LECA

Gaspard Dussert, Vincent Miele, Simon Chamaillé-Jammes, Stéphane Dray

@misc{dussert2025attention,
    author = {Dussert, Gaspard and Dray, Stéphane and Chamaillé-Jammes, Simon and Miele, Vincent},
    title = {Paying Attention to Other Animal Detections Improves Camera Trap Classification Models},
    year = {2025},
    doi = {10.1101/2025.07.15.664849},
    publisher = {bioRxiv},
    type = {preprint},
}

Overview

Multi-Crop Attention-based Classifier uses self-attention to improve classification by levaring the context of other detection within the camera trap sequence.

GitHub repo is accessible here : https://github.com/gdussert/MCA_Classifier

Note: Due to the 1GB storage constraint on Hugging Face, the crop_classifier has been converted to float16.

Acknowledgements

This work was granted access to the HPC resources of IDRIS under the allocation 2022-AD010113729 made by GENCI.

Thanks to LILA BC for hosting the two datasets, Snapshot Safari 2024 Expansion and Snapshot Serengeti, and for all those who contributed to it.

Thanks MegaDetector and PytorchWildlife: MDv5a was used for preprocessing and MDV6-Ultralytics-YoloV10-Extra is used by the Gradio demo.

Thanks to the amazing timm repository.