Online Pseudo-Label Unified Object Detection for Multiple Datasets Training
It addresses the problem of comprehensive object detection across multiple datasets for researchers and practitioners, but is incremental as it builds on existing UOD frameworks.
The paper tackles the cross-dataset missing annotations issue in Unified Object Detection by proposing an online pseudo-label scheme that uses a periodically updated teacher model to generate pseudo-labels and improves box regression with category-specific methods, achieving higher accuracy than existing SOTA methods on benchmarks like COCO, Object365, and OpenImages.
The Unified Object Detection (UOD) task aims to achieve object detection of all merged categories through training on multiple datasets, and is of great significance in comprehensive object detection scenarios. In this paper, we conduct a thorough analysis of the cross datasets missing annotations issue, and propose an Online Pseudo-Label Unified Object Detection scheme. Our method uses a periodically updated teacher model to generate pseudo-labels for the unlabelled objects in each sub-dataset. This periodical update strategy could better ensure that the accuracy of the teacher model reaches the local maxima and maximized the quality of pseudo-labels. In addition, we survey the influence of overlapped region proposals on the accuracy of box regression. We propose a category specific box regression and a pseudo-label RPN head to improve the recall rate of the Region Proposal Network (PRN). Our experimental results on common used benchmarks (\eg COCO, Object365 and OpenImages) indicates that our online pseudo-label UOD method achieves higher accuracy than existing SOTA methods.