FEW-Shot Object Detection with Foreground Augment and Background Attenuation
Ying Zeng, Haoliang Yuan · 2022
Few-shot object detection has made prominent progress in the development of object detection tasks owning to its ability of detection under extremely few annotated data. In this paper, we put forth a novel few shot object detection method which consists of foreground augment and background attenuation module. This approach is proposed to alleviate the impact of irrelevant contextual information in novel categories. It has proven to be a powerful foreground augment and background attenuation framework for few-shot object detection. Comprehensive experiments on PASCAL VOC benchmarks demonstrate the effectiveness of our approach.