Incremental multi-target domain adaptation for object detection with efficient domain transfer
作者：
Highlights：
• Unsupervised multi-target domain adaptation for object detection.
• Formulate multi-target adaptation problem as an incremental adaptation problem.
• Take advantages of incremental learning to avoid retain/retrain on previous data.
• Achieve high level of accuracy across multiple target domains/datasets.
• The proposed method is also efficient in term of adaptation complexity.
摘要
•Unsupervised multi-target domain adaptation for object detection.•Formulate multi-target adaptation problem as an incremental adaptation problem.•Take advantages of incremental learning to avoid retain/retrain on previous data.•Achieve high level of accuracy across multiple target domains/datasets.•The proposed method is also efficient in term of adaptation complexity.
论文关键词：Deep learning,Convolutional NNs,Object detection,Unsupervised domain adaptation,Multi-Target domain adaptation,Incremental learning
论文评审过程：Received 10 August 2021, Revised 26 April 2022, Accepted 30 April 2022, Available online 3 May 2022, Version of Record 8 May 2022.
论文官网地址：https://doi.org/10.1016/j.patcog.2022.108771
