Dynamic Head: Unifying Object Detection Heads with Attentions
The complex nature of combining localization and classification in object detection has resulted in the flourished development of methods. Previous works tried to improve the performance in various object detection heads but failed to present a unified view.
Also cited · not yet reviewed (3)
- Faster R-CNN2015 · cited 4×, 1 in Method“Two-stage detectors utilize region proposal and ROI-pooling [23] layers to extract intermediate representations from feature pyramid of a backbone network.”From this paper · §Our Approach
- Transformer2017 · cited 1×, 1 in Method“Recently, there is a trend to introduce the Transformer module [29] from natural language processing into computer vision tasks.”From this paper · §Our Approach
- Swin2021 · cited 3דWhen training our dynamic head with latest transformer backbone [19], extra data and increased input size, we can further improve the current SOTA on COCO benchmark.”From this paper · §Appendix
Led to
- Florence2021 · cited 5×, 2 in Method“For this goal, we add an adaptor Dynamic Head (Dai et al. 2021a) (or Dynamic DETR (Dai et al. 2021b)), a unified attention mechanism for the detection head, to the pretrained image encoder (i.e. , CoSwin).”From Florence · §Approach
Abstract
The complex nature of combining localization and classification in object detection has resulted in the flourished development of methods. Previous works tried to improve the performance in various object detection heads but failed to present a unified view. In this paper, we present a novel dynamic head framework to unify object detection heads with attentions. By coherently combining multiple self-attention mechanisms between feature levels for scale-awareness, among spatial locations for spatial-awareness, and within output channels for task-awareness, the proposed approach significantly improves the representation ability of object detection heads without any computational overhead. Further experiments demonstrate that the effectiveness and efficiency of the proposed dynamic head on the COCO benchmark. With a standard ResNeXt-101-DCN backbone, we largely improve the performance over popular object detectors and achieve a new state-of-the-art at 54.0 AP. Furthermore, with latest transformer backbone and extra data, we can push current best COCO result to a new record at 60.6 AP. The code will be released at https://github.com/microsoft/DynamicHead.