Multi-Posture Human Extraction Using Coarse-to-Fine Method and Human Skeleton Method

摘要

Human detection is a technology that is used in various elds. It is, however, a trade-off between detectionaccuracy and precise extraction of human regions. The purpose of this paper is, therefore, to extract a precisehuman region using Coarse-to-Fine Method or Human Skeleton Method. In the Coarse-to-Fine Method, atrst, Coarse Detector detects multi-posture humans. Next, Fine Detector extracts precise human regions. Inthe Human Skeleton Method, human skeletons are extracted by using OpenPose~1 . Next, skeleton images aredilated based on physique of the human. Finally, human regions are extracted by using GrabCut~2 . Theextracted results are evaluated by detection accuracy and preciseness of extracted regions. F-measure and IoU(Intersection over Union) are employed to evaluate the detection accuracy and the preciseness respectively. Inthe Coarse-to-Fine Method, the detection accuracy is 0.523, and the extraction preciseness becomes 0.807. Inthe Human Skeleton Method, the detection accuracy reaches 0.928, and the extraction preciseness is 0.868.Especially, Human Skeleton Method gets excellent performance in not only the extraction preciseness but alsothe detection accuracy.