pycls is an image classification codebase, written in PyTorch. The codebase was originally developed for a project that led to the On Network Design Spaces for Visual Recognition work. pycls has since matured into a general image classification codebase that has been adopted by a number representation learning projects at Facebook AI Research.
The goal of pycls is to provide a high-quality, high-performance codebase for image classification research. It is designed to be simple and flexible in order to support rapid implementation and evaluation of research ideas.
The codebase implements efficient single-machine multi-gpu training, powered by PyTorch distributed package. pycls includes implementations of standard baseline models (ResNet, ResNeXt, EfficientNet) and generic modeling functionality that can be useful for experimenting with network design. Additional models can be easily implemented.
Please see INSTALL.md for installation instructions.
After installation, please see GETTING_STARTED.md for basic instructions on training and evaluation with pycls.
Coming soon!
If you use pycls in your research, please use the following BibTex entry
@InProceedings{Radosavovic2019,
title = {On Network Design Spaces for Visual Recognition},
author = {Radosavovic, Ilija and Johnson, Justin and Xie, Saining and Lo, Wan-Yen and Doll{\'a}r, Piotr},
booktitle = {ICCV},
year = {2019},
}
pycls is released under the MIT license. See the LICENSE file for more information.