Deformable DETR: Deformable Transformers for End-to-End Object Detection
This addresses efficiency and accuracy issues in object detection for computer vision applications, representing a significant incremental improvement.
The paper tackled DETR's slow convergence and limited feature resolution in object detection by proposing Deformable DETR, which uses attention modules focusing on key sampling points, achieving better performance than DETR with 10 times fewer training epochs.
DETR has been recently proposed to eliminate the need for many hand-designed components in object detection while demonstrating good performance. However, it suffers from slow convergence and limited feature spatial resolution, due to the limitation of Transformer attention modules in processing image feature maps. To mitigate these issues, we proposed Deformable DETR, whose attention modules only attend to a small set of key sampling points around a reference. Deformable DETR can achieve better performance than DETR (especially on small objects) with 10 times less training epochs. Extensive experiments on the COCO benchmark demonstrate the effectiveness of our approach. Code is released at https://github.com/fundamentalvision/Deformable-DETR.