SCL: Towards Accurate Domain Adaptive Object Detection via Gradient Detach Based Stacked Complementary Losses

Abstract

Unsupervised domain adaptive object detection aims to learn a robust detector in the domain shift circumstance, where the training (source) domain is label-rich with bounding box annotations, while the testing (target) domain is label-agnostic and the feature distributions between training and testing domains are dissimilar or even totally different. In this paper, we propose a gradient detach based stacked complementary losses (SCL) method that uses detection losses as the primary objective, and cuts in several auxiliary losses in different network stages accompanying with gradient detach training to learn more discriminative representations. We argue that the prior methods mainly leverage more loss functions for training but ignore the interaction of different losses and also the compatible training strategy (gradient detach updating in our work). Thus, our proposed method is a more syncretic adaptation learning process. We conduct comprehensive experiments on seven datasets, the results demonstrate that our method performs favorably better than the state-of-the-art methods by a significant margin. For instance, from Cityscapes to FoggyCityscapes, we achieve 37.9% mAP, outperforming the previous art Strong-Weak by 3.6%.

References in corpus (5)

• Semantic Foggy Scene Understanding with Synthetic Data
• Unsupervised Domain Adaptation with Residual Transfer Networks
• On Learning Invariant Representation for Domain Adaptation
• Complement Objective Training
• SPLAT: Semantic Pixel-Level Adaptation Transforms for Detection

Cited by in corpus (15)

• PIT: Position-Invariant Transform for Cross-FoV Domain Adaptation
• Decoupled Adaptation for Cross-Domain Object Detection
• Cross-Domain Object Detection Using Unsupervised Image Translation
• Unbiased Mean Teacher for Cross-domain Object Detection
• Cross-domain Object Detection through Coarse-to-Fine Feature Adaptation
• Domain Adaptive YOLO for One-Stage Cross-Domain Detection
• Domain Adaptive Object Detection via Asymmetric Tri-way Faster-RCNN
• Deep Domain Adaptive Object Detection: a Survey
• Self-Adversarial Disentangling for Specific Domain Adaptation
• SF-UDA${}^{3D}$: Source-Free Unsupervised Domain Adaptation for LiDAR-Based 3D Object Detection
• A Survey on Deep Domain Adaptation and Tiny Object Detection Challenges, Techniques and Datasets
• Spatial Attention Pyramid Network for Unsupervised Domain Adaptation
• Adaptive Object Detection with Dual Multi-Label Prediction
• A Pairwise DomMix Attentive Adversarial Network for Unsupervised Domain Adaptive Object Detection
• Channel-wise Alignment for Adaptive Object Detection