DETR
End-to-end object detection with transformers by Meta, eliminating hand-designed components.
About
DETR, the Detection Transformer from Meta, reframes object detection as a direct set-prediction problem solved by a transformer encoder-decoder, removing hand-designed components like anchor boxes and non-maximum suppression. A bipartite matching loss forces unique predictions, and the approach matches a Faster R-CNN ResNet-50 baseline on COCO with simpler inference code. Pretrained models are provided. Released under the Apache 2.0 license.
Reviews (0)
Leave a Review
No reviews yet. Be the first to review!
Details
- Price
- Free
- Platform
- Local/Desktop
- Difficulty
- Advanced (4/5)
- License
- Apache-2.0
- Minimum VRAM
- 8 GB
- Added
- Apr 3, 2026
Related Tools
Foundation model for monocular depth estimation by TikTok.
Improved vision-language model by Google using sigmoid loss for contrastive learning.
Open-vocabulary object detection model by Google using vision transformers.
OpenMMLab detection toolbox with 300+ pre-trained models and 80+ algorithms.
State-of-the-art real-time object detection supporting YOLOv5 through v11.
Additive angular margin loss for deep face recognition.