Dually Supervised Feature Pyramid for Object Detection and Segmentation
Fan Yang,Cheng Lu,Yandong Guo,Longin Jan Latecki,Haibin Ling
DOI: https://doi.org/10.48550/arXiv.1912.03730
2019-12-14
Abstract:Feature pyramid architecture has been broadly adopted in object detection and segmentation to deal with multi-scale problem. However, in this paper we show that the capacity of the architecture has not been fully explored due to the inadequate utilization of the supervision information. Such insufficient utilization is caused by the supervision signal degradation in back propagation. Thus inspired, we propose a dually supervised method, named dually supervised FPN (DSFPN), to enhance the supervision signal when training the feature pyramid network (FPN). In particular, DSFPN is constructed by attaching extra prediction (i.e., detection or segmentation) heads to the bottom-up subnet of FPN. Hence, the features can be optimized by the additional heads before being forwarded to subsequent networks. Further, the auxiliary heads can serve as a regularization term to facilitate the model training. In addition, to strengthen the capability of the detection heads in DSFPN for handling two inhomogeneous tasks, i.e., classification and regression, the originally shared hidden feature space is separated by decoupling classification and regression subnets. To demonstrate the generalizability, effectiveness, and efficiency of the proposed method, DSFPN is integrated into four representative detectors (Faster RCNN, Mask RCNN, Cascade RCNN, and Cascade Mask RCNN) and assessed on the MS COCO dataset. Promising precision improvement, state-of-the-art performance, and negligible additional computational cost are demonstrated through extensive experiments. Code will be provided.
Computer Vision and Pattern Recognition