Real Time Crowd Counting: A Review
Salwa Thasveen M., L. Mredhula · 2020
Crowd counting is a process of counting number of people or objects in videos or images. This process has various applications related to our day to day life such as urban planning, health care, disaster management, public safety management, and defense. Thus new researches are going on in this field. The crowd techniques are broadly classified as supervised learning based and unsupervised learning based techniques. Some of traditional and convolutional neural network crowd counting techniques are discussed in this paper. Crowd counting techniques are facing various limitations such as occlusion, distortion in scale and perspective, non uniform distribution. As the crowd density increases, calculation complexity also increases. Most of the crowd counting methods involves density estimation. Density estimation gives an idea about the spatial distribution of people along with the count.