DINO (DETR)

library paper
IDEA Lab
DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection. Uses contrastive denoising training, look-forward-twice box prediction, and mixed query selection. Achieves 63.3 AP on COCO test-dev with SwinL.
Paper (arXiv)GitHub

Outputs 2

DINO (DETR)

library

Official implementation of DINO end-to-end object detection. 49.4 AP in 12 epochs and 51.3 AP in 24 epochs with ResNet-50.

GitHub

GitHub Repository

DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection

paper

Introduces contrastive denoising, look-forward-twice, and mixed query selection for DETR-based detection.

Paper (arXiv)

arXiv: 2203.03605

Venue: ICLR 2023

visionopen-source