Single-Shot Refinement Neural Network for Object Detection

For object detection, the two-stage approach (e.g., Faster R-CNN) has been achieving the highest accuracy, whereas the one-stage approach (e.g., SSD) has the advantage of high efficiency. To inherit the merits of both while overcoming their disadvantages, in this paper, we propose a novel single-shot based detector, called RefineDet, that achieves better accuracy than two-stage methods and maintains comparable efficiency of one-stage methods. RefineDet consists of two inter-connected modules, namely, the anchor refinement module and the object detection module. Specifically, the former aims to (1) filter out negative anchors to reduce search space for the classifier, and (2) coarsely adjust the locations and sizes of anchors to provide better initialization for the subsequent regressor. The latter module takes the refined anchors as the input from the former to further improve the regression and predict multi-class label. Meanwhile, we design a transfer connection block to transfer the features in the anchor refinement module to predict locations, sizes and class labels of objects in the object detection module. The multi-task loss function enables us to train the whole network in an end-to-end way. Extensive experiments on PASCAL VOC 2007, PASCAL VOC 2012, and MS COCO demonstrate that RefineDet achieves state-of-the-art detection accuracy with high efficiency. Code is available at https://github.com/sfzhang15/RefineDet

PDF Abstract CVPR 2018 PDF CVPR 2018 Abstract

Datasets


Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Object Detection COCO test-dev RefineDet512 (ResNet-101) box mAP 36.4 # 218
AP50 57.5 # 149
AP75 39.5 # 150
APS 16.6 # 141
APM 39.9 # 139
APL 51.4 # 127
Hardware Burden None # 1
Operations per network pass None # 1
Object Detection COCO test-dev RefineDet512+ (ResNet-101) box mAP 41.8 # 175
AP50 62.9 # 104
AP75 45.7 # 119
APS 25.6 # 91
APM 45.1 # 109
APL 54.1 # 109
Hardware Burden None # 1
Operations per network pass None # 1
Object Detection COCO test-dev RefineDet512 (VGG-16) box mAP 33 # 229
AP50 54.5 # 153
AP75 35.5 # 152
APS 16.3 # 142
APM 36.3 # 142
APL 44.3 # 144
Hardware Burden None # 1
Operations per network pass None # 1
Object Detection COCO test-dev RefineDet512+ (VGG-16) box mAP 37.6 # 212
AP50 58.7 # 143
AP75 40.8 # 147
APS 22.7 # 118
APM 40.3 # 137
APL 48.3 # 141
Hardware Burden None # 1
Operations per network pass None # 1

Methods