结合密集连接的轻量级高分辨率人体姿态估计  

Lightweight high-resolution human pose estimation combined with densely connected network

作　　者：高坤 李汪根 束阳 葛英奎 王志格 Gao Kun;Li Wanggen;Shu Yang;Ge Yingkui;Wang Zhige(School of Computer and Information,Anhui Normal University,Wuhu 241002,China)

机构地区：[1]安徽师范大学计算机与信息学院,芜湖241002

出　　处：《中国图象图形学报》2024年第5期1408-1420,共13页Journal of Image and Graphics

基　　金：国家自然科学基金项目(61976006)。

摘　　要：目的 为了更好地实现轻量化的人体姿态估计,在轻量级模型极为有限的资源下实现更高的检测性能。基于高分辨率网络(high resolution network,HRNet)提出了结合密集连接网络的轻量级高分辨率人体姿态估计网络(lightweight high-resolution human estimation combined with densely connected network,LDHNet)。方法 通过重新设计HRNet中的阶段分支结构以及提出新的轻量级特征提取模块,构建了轻量高效的特征提取单元,同时对多分支之间特征融合部分进行了轻量化改进,进一步降低模型的复杂度,最终大幅降低了模型的参数量与计算量,实现了轻量化的设计目标,并且保证了模型的性能。结果 实验表明,在MPII(Max Planck Institute for Informatics)测试集上相比于自顶向下的轻量级人体姿态估计模型LiteHRNet,LDHNet仅通过增加少量参数量与计算量,平均预测准确度即提升了1.5%,与LiteHRNet的改进型DiteHRNet相比也提升了0.9%,在COCO(common objects in context)验证集上的结果表明,与LiteHRNet相比,LDHNet的平均检测准确度提升了3.4%,与DiteHRNet相比也提升了2.3%,与融合Transformer的HRFormer相比,LDHNet在参数量和计算量都更低的条件下有近似的检测性能,在面对实际场景时LDHNet也有着稳定的表现,在同样的环境下LDHNet的推理速度要高于基线HRNet以及LiteHRNet等。结论 该模型有效实现了轻量化并保证了预测性能。Objective Human pose estimation is a technology that can be widely used in life.In recent years,many excel⁃lent high-precision methods have been proposed,but they are often accompanied by a very large model scale,which will encounter the problem of computing power bottleneck in application.Whether for model training or deployment,large mod⁃els require considerable computing power as the basis.Most of them have low computing power.Similarly,for the scenes in daily life,the equipment needs further applicability and detection speed of the model,which is difficult to achieve by large models.Given such requirements,lightweight human pose estimation has become a hot research field.The main problem is how to achieve high detection accuracy and fast detection speed under the extremely limited number of resources.Lightweight models will inevitably fall into a disadvantage in detection accuracy compared with large models.However, fortunately, from many studies in recent years, the lightweight model can also achieve higher detection accuracythan large ones. A good balance can be reached between them. Method Based on a high-resolution network (HRNet), alightweight high-resolution human pose estimation network combined with a dense connection network (LDHNet) was pro⁃posed. First, dense connection and multi-scale were integrated to construct a lightweight and efficient feature extractionunit by redesigning the stage branch structure in HRNet and proposing a new lightweight feature extraction module. Then,the feature extraction module is composed of modules similar to the pyramid structure, and the dilated convolution of threescales is used to obtain a wide range of feature information in the feature map by stacking the multi-layer feature extractionmodules and fusing the output of each layer. The concatenation of the output feature map of the feature extraction module isto reuse the feature map and fully extract the information contained in the feature map. These two points can make up forthe problem of insufficient uti

关 键 词：人体姿态估计 轻量级网络 密集连接网络 高分辨率网络 多分支结构 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]