Fast Convergence of DETR with Spatially Modulated Co-Attention

• Main Authors: Gao, Peng, Zheng, Minghang, Wang, Xiaogang, Dai, Jifeng, Li, Hongsheng
• Format: Conference Proceeding
• Language: English
• Subjects: Costs; Detection and localization in 2D and 3D; Object detection; Representation learning; Schedules; Training; Transformers; Vision applications and systems; Visualization
• ISSN: 2380-7504
• DOI: 10.1109/ICCV48922.2021.00360

The recently proposed Detection Transformer (DETR) model successfully applies Transformer to objects detection and achieves comparable performance with two-stage object detection frameworks, such as Faster-RCNN. However, DETR suffers from its slow convergence. Training DETR [4] from scratch needs 500 epochs to achieve a high accuracy. To accelerate its convergence, we propose a simple yet effective scheme for improving the DETR framework, namely Spatially Modulated Co-Attention (SMCA) mechanism. The core idea of SMCA is to conduct location-aware co-attention in DETR by constraining co-attention responses to be high near initially estimated bounding box locations. Our proposed SMCA increases DETR's convergence speed by replacing the original co-attention mechanism in the decoder while keeping other operations in DETR unchanged. Furthermore, by integrating multi-head and scale-selection attention designs into SMCA, our fully-fledged SMCA can achieve better performance compared to DETR with a dilated convolution-based backbone (45.6 mAP at 108 epochs vs. 43.3 mAP at 500 epochs). We perform extensive ablation studies on COCO dataset to validate SMCA. Code is released at https://github.com/gaopengcuhk/SMCA-DETR.