ZhangGongjie/Meta-DETR
A PyTorch implementation of Meta-DETR, a meta-learning-based few-shot object detector published in IEEE T-PAMI 2022.
Not currently ranked — collecting fresh signals.
star history
Meta-DETR performs image-level few-shot object detection by directly predicting bounding boxes without region proposal networks, bypassing the proposal quality gap that limits R-CNN-based detectors. It exploits inter-class correlation to improve generalization from base to novel classes. The model uses meta-learning on object queries rather than region proposals to achieve state-of-the-art few-shot detection performance.
Frequently asked
- What is ZhangGongjie/Meta-DETR?
- A PyTorch implementation of Meta-DETR, a meta-learning-based few-shot object detector published in IEEE T-PAMI 2022.
- Is Meta-DETR open source?
- Yes — ZhangGongjie/Meta-DETR is open source, released under the MIT license.
- What language is Meta-DETR written in?
- ZhangGongjie/Meta-DETR is primarily written in Python.
- How popular is Meta-DETR?
- ZhangGongjie/Meta-DETR has 435 stars on GitHub.
- Where can I find Meta-DETR?
- ZhangGongjie/Meta-DETR is on GitHub at https://github.com/ZhangGongjie/Meta-DETR.