Many-shot from Low-shot: Learning to Annotate using Mixed Supervision for Object Detection

Object detection has witnessed significant progress by relying on large,\nmanually annotated datasets. Annotating such datasets is highly time consuming\nand expensive, which motivates the development of weakly supervised and\nfew-shot object detection methods. However, these methods largely underperform\nwith respect to their strongly supervised counterpart, as weak training signals\n\\emph{often} result in partial or oversized detections. Towards solving this\nproblem we introduce, for the first time, an online annotation module (OAM)\nthat learns to generate a many-shot set of \\emph{reliable} annotations from a\nlarger volume of weakly labelled images. Our OAM can be jointly trained with\nany fully supervised two-stage object detection method, providing additional\ntraining annotations on the fly. This results in a fully end-to-end strategy\nthat only requires a low-shot set of fully annotated images. The integration of\nthe OAM with Fast(er) R-CNN improves their performance by $17\\%$ mAP, $9\\%$\nAP50 on PASCAL VOC 2007 and MS-COCO benchmarks, and significantly outperforms\ncompeting methods using mixed supervision.\n