人体解析针对图像中人体不同部位进行语义分割，是近年来计算机视觉领域中的一个重要研究课题。不同于场景中的一般物体，人具有高度的结构化特征，并存在复杂的姿态变化和衣物遮挡情况。针对这一任务，提出一种基于卷积神经网络的层次化标签结构的人体解析方法。首先对精细的标签按照类别进行不同程度的合并，获得多个层级的解析图；然后改进具有金字塔特征抽取结构的卷积神经网络，使用解析图对金字塔不同层级的特征进行监督；最后将所有层级特征进行融合得到解析结果。在人体解析数据集LIP上的实验验证，与当前通用的语义分割算法相比，该算法可获得更高的人体解析准确性并改善了图像的分割效果。
Human parsing is a type of semantic segmentation of different human body parts in an image. It is an emerging task in the field of computer vision. Compared with general objects, human body is much more structured but with wide variations in pose and occlusions caused by wearing. In this paper we present a hierarchical label network (HLNet). Firstly, fine categories are merged into body parts with different granularities to obtain multiple parsing maps for each image. Next, a convolutional neural network with a pyramid feature extraction structure is trained under supervision of these maps. Finally, the hierarchical features are fused together to predict the final parsing results. Experimental results on the LIP dataset show that the proposed algorithm achieves higher parsing accuracy and better segmentation performance, compared with common semantic segmentation algorithms.
[1] Yue-Hei Ng J, Hausknecht M, Vijayanarasimhan S, et al. Beyond short snippets:deep networks for video classification[C]//Computer Vision & Pattern Recognition. IEEE, 2015:4694-4702.
[2] Lin D, Lu C, Liao R, et al. Learning important spatial pooling regions for scene classification[C]//Computer Vision & Pattern Recognition. IEEE, 2014:3726-3733.
[3] Smolyanskiy N, Kamenev A, Birchfield S. On the importance of stereo for accurate depth estimation:an efficient semi-supervised deep neural network approach[C]//Computer Vision & Pattern Recognition Workshops. IEEE, 2018:1007-1015.
[4] Krizhevsky A, Sutskever I, Hinton G E. ImageNet classification with deep convolutional neural networks[C]//Advances in Neural Information Processing Systems. IEEE, 2012:1097-1105.
[5] Ren S, He K, Girshick R, et al. Faster R-CNN:towards real-time object detection with region proposal networks[C]//Advances in Neural Information Processing Systems. IEEE, 2015:91-99.
[6] Chen L C, Papandreou G, Kokkinos I, et al. Deeplab:semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2018, 40(4):834-848.
[7] Bourdev L, Malik J. Poselets:body part detectors trained using 3d human pose annotations[C]//International Conference on Computer Vision. IEEE, 2009:1365-1372.
[8] Bo Y, Fowlkes C C. Shape-based pedestrian parsing[C]//Computer Vision & Pattern Recognition. IEEE, 2011:2265-2272.
[9] Yamaguchi K, Kiapour M H, Ortiz L E, et al. Parsing clothing in fashion photographs[C]//Proceedings of the 2010 IEEE Conference on Computer Vision & Pattern Recognition. IEEE, 2012:3570-3577.
[10] Rauschert I, Collins R T. A generative model for simultaneous estimation of human body shape and pixel-level segmentation[C]//European Conference on Computer Vision. IEEE, 2012:704-717.
[11] Dong J, Chen Q, Xia W, et al. A deformable mixture parsing model with parselets[C]//Computer Vision & Pattern Recognition. IEEE, 2013:3408-3415.
[12] Liang X, Gong K, Shen X, et al. Look into person:joint body parsing & pose estimation network and a new benchmark[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2018, 41(4):871-885.
[13] Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2014, 39(4):640-651.
[14] Chen L C, Papandreou G, Schroff F, et al. Rethinking atrous convolution for semantic image segmentation[C]//arXiv preprint, arXiv, 2017:1706.05587.
[15] Chen L C, Yang Y, Wang J, et al. Attention to scale:scale-aware semantic image segmentation[C]//Computer Vision & Pattern Recognition. IEEE, 2016:3640-3649.
[16] Nie X, Feng J, Yan S. Mutual learning to adapt for joint human parsing and pose estimation[C]//European Conference on Computer Vision. IEEE, 2018:502-517.
[17] Luo Y, Zheng Z, Zheng L, et al. Macro-micro adversarial network for human parsing[C]//European Conference on Computer Vision. IEEE, 2018:418-434.
[18] Zhao J, Li J, Nie X, et al. Self-supervised neural aggregation networks for human parsing[C]//Computer Vision & Pattern Recognition Workshops. IEEE, 2017:7-15.
[19] Liu T, Ruan T, Huang Z, et al. Devil in the details:towards accurate single and multiple human parsing[C]//arXiv preprint, arXiv, 2018:1809.05996.
[20] Zhao H, Shi J, Qi X, et al. Pyramid scene parsing network[C]//IEEE Conference on Computer Vision & Pattern Recognition. IEEE, 2017:2881-2890.
