IDEA-Research/DN-DETR
DN-DETR is a transformer-based object detection model that accelerates DETR training through query denoising, achieving CVPR 2022 Oral status.
Not currently ranked — collecting fresh signals.
star history
DN-DETR (DeNoising DETR) is an official implementation of a CVPR 2022 Oral paper that introduces query denoising to accelerate DETR (DEtection TRansformer) training. The method applies denoising techniques to transformer queries during training to improve convergence speed and detection accuracy. The repository includes training code, model weights, and extensions to other DETR variants including Mask2Former and Faster R-CNN.
Frequently asked
- What is IDEA-Research/DN-DETR?
- DN-DETR is a transformer-based object detection model that accelerates DETR training through query denoising, achieving CVPR 2022 Oral status.
- Is DN-DETR open source?
- Yes — IDEA-Research/DN-DETR is open source, released under the Apache-2.0 license.
- What language is DN-DETR written in?
- IDEA-Research/DN-DETR is primarily written in Python.
- How popular is DN-DETR?
- IDEA-Research/DN-DETR has 605 stars on GitHub.
- Where can I find DN-DETR?
- IDEA-Research/DN-DETR is on GitHub at https://github.com/IDEA-Research/DN-DETR.