Happy to share our new open-source work: detrex. detrex is an open-source toolbox that provides state-of-the-art Transformer-based object detection algorithms.

Here's the github links: https://github.com/IDEA-Research/detrex

detrex was built on detectron2 and we used the powerful LazyConfig system for more flexible syntax and cleaner config files.

Fun Facts:

Its name says all about this work:

• detr-ex: We take our hats off to DETR and regard this repo as an extension of Transformer-based detection algorithms.

• det-rex: rex literally means 'king' in Latin. We hope this repo can help advance the state of the art on object detection by providing the best Transformer-based detection algorithms from the research community.

• de-t.rex: de means 'the' in German. T.rex, also called Tyrannosaurus Rex, means 'king of the tyrant lizards' and connects to our research work 'DINO', which is short for Dinosaur.

Overview of Model Zoo

detrex has already implemented the following algorithms:

• DETR (ECCV'2020)
• Deformable-DETR (ICLR'2021 Oral)
• Conditional-DETR (ICCV'2021)
• DAB-DETR (ICLR'2022)
• DAB-Deformable-DETR (ICLR'2022)
• DN-DETR (CVPR'2022 Oral)
• DN-Deformable-DETR (CVPR'2022 Oral)
• DINO (ArXiv'2022)
• Group-DETR (ArXiv' 2022)

Nice Performance

Here is the performance of models trained using detrex compared with their original repo.