This repository is the official implementation of the Battle of the Backbones: A Large-Scale Comparison of Pretrained Models across Computer Vision Tasks.
Battle of the Backbones (BoB) compares many popular publicly available pretrained checkpoints, as well as randomly initialized baselines, on a wide variety of downstream tasks including image classification on natural, medical, and satellite images, object detection and segmentation, out-of-distribution generalization, and image retrieval.
For each vision task, please refer to its respective repository for detailed documentation, source code, and further information:
🌟 Image Classification: BoB-Classification
🌟 Object Detection and Segmentation: BoB-Detection
🌟 Out-of-Distribution Image Classification: BoB-OOD-Classification
🌟 Out-of-Distribution Object Detection: BoB-OOD-Detection
🌟 Image Retrieval: BoB-Retrieval
If you find this useful in your research, please cite our paper:
@misc{goldblum2023battle,
title={Battle of the Backbones: A Large-Scale Comparison of Pretrained Models across Computer Vision Tasks},
author={Micah Goldblum and Hossein Souri and Renkun Ni and Manli Shu and Viraj Prabhu and Gowthami Somepalli and Prithvijit Chattopadhyay and Mark Ibrahim and Adrien Bardes and Judy Hoffman and Rama Chellappa and Andrew Gordon Wilson and Tom Goldstein},
year={2023},
eprint={2310.19909},
archivePrefix={arXiv},
primaryClass={cs.CV}
}