Improvement of Classification in One-Stage Detector
This work provides an incremental improvement to the classification performance of one-stage object detectors, which is beneficial for researchers and practitioners working on real-time object detection.
This paper addresses the misalignment between classification and localization in one-stage detectors, specifically RetinaNet, where many well-localized boxes have low classification scores. By introducing an object confidence task that uses IoU as targets and focuses on positive samples, the method improves classification and eases misalignment, achieving AP gains of 0.7% with ResNet50 and 1.0% with ResNet101 on MS COCO 2017.
RetinaNet proposed Focal Loss for classification task and improved one-stage detectors greatly. However, there is still a gap between it and two-stage detectors. We analyze the prediction of RetinaNet and find that the misalignment of classification and localization is the main factor. Most of predicted boxes, whose IoU with ground-truth boxes are greater than 0.5, while their classification scores are lower than 0.5, which shows that the classification task still needs to be optimized. In this paper we proposed an object confidence task for this problem, and it shares features with classification task. This task uses IoUs between samples and ground-truth boxes as targets, and it only uses losses of positive samples in training, which can increase loss weight of positive samples in classification task training. Also the joint of classification score and object confidence will be used to guide NMS. Our method can not only improve classification task, but also ease misalignment of classification and localization. To evaluate the effectiveness of this method, we show our experiments on MS COCO 2017 dataset. Without whistles and bells, our method can improve AP by 0.7% and 1.0% on COCO validation dataset with ResNet50 and ResNet101 respectively at same training configs, and it can achieve 38.4% AP with two times training time. Code is at: http://github.com/chenzuge1/RetinaNet-Conf.git.