基于深度学习的图像人群计数方法浅析

从人群图像中估算人的数量有着广泛的应用,如视频监控、公共安全以及城市规划等。近年来,深度学习方法的使用为遮挡、视角变化等挑战带来了较好的解决方案。本文首先回顾了传统人群计数的方法,然后对基于深度学习的方法做了浅析,最后探讨了该技术的发展方向。

Estimating the number of people from crowd images has a wide range of applications, such as video monitoring, public safety and urban planning. In recent years, the use of deep learning brings better solutions to the challenges such as occlusion and perspective changes. In this paper, traditional crowd counting methods are reviewed firstly, then a brief analysis of deep learning-based methods are made, and promising avenues of research in this field are identified at last.