Instance Localization for Self-supervised Detection Pretraining
This addresses the specific problem of improving self-supervised pretraining for object detection, which is incremental as it builds on prior self-supervised learning methods.
The paper tackled the problem of degraded transfer performance of self-supervised learning from image classification to object detection by proposing a new pretext task called instance localization, which integrates bounding boxes into pretraining and yields state-of-the-art results on PASCAL VOC and MSCOCO.
Prior research on self-supervised learning has led to considerable progress on image classification, but often with degraded transfer performance on object detection. The objective of this paper is to advance self-supervised pretrained models specifically for object detection. Based on the inherent difference between classification and detection, we propose a new self-supervised pretext task, called instance localization. Image instances are pasted at various locations and scales onto background images. The pretext task is to predict the instance category given the composited images as well as the foreground bounding boxes. We show that integration of bounding boxes into pretraining promotes better task alignment and architecture alignment for transfer learning. In addition, we propose an augmentation method on the bounding boxes to further enhance the feature alignment. As a result, our model becomes weaker at Imagenet semantic classification but stronger at image patch localization, with an overall stronger pretrained model for object detection. Experimental results demonstrate that our approach yields state-of-the-art transfer learning results for object detection on PASCAL VOC and MSCOCO.