基于部件精细化分割的行人检索方法

Person Retrieval Based on Fine-Part Segmentation

由于摄像机的拍摄角度和清晰度等存在差异,行人图像存在场景变化、走路姿态改变等区别,导致行人检索存在精度低的问题,由此提出一种对部件进行精细分割的行人检索方法。通过精细化分割部件,提升部件内部的一致性,提高检索的精度。在-1501和DukemMTMC-reID两个数据集上进行实验,结果表明,该方法优于传统的部件分割方法。

Because of the camera shooting Angle and clarity and other differences.There are differences in pedestrian image such as scene change and walking posture change.Pedestrian retrieval has the problem of low precision.A pedestrian retrieval method for fine segmentation of parts is proposed.By finely segmented parts.The internal consistency of components is improved and the retrieval precision is improved.Experiments were carried out on Market-1501 and DukemMTMC-reID data sets in this paper,and the results show that this method is superior to the traditional part segmentation method.