A Survey of Crowd Counting CNN-based

Abstract

Benefiting from the powerful feature representation ability of deep learning, Convolutional Neural Network (CNN) provides a better solution to estimate accurately the number of people in a crowded scene, but it still faces many problems that need to be solved urgently. It is one of the key and difficult points in the field to reduce the complexity of the network and to improve the real-time performance of the network, so as to improve the accuracy of crowd counting. Firstly, this paper introduces the research background and application of crowd counting. Secondly, it focuses on the commonly used counting model, loss function, and dataset and evaluation method. Then compare the performance structure, advantages and disadvantages of different algorithms horizontally on several published datasets. Finally, it summarizes the shortcomings of the existing crowd counting, put forward to the future research direction of crowd counting.

GhostCount: A lightweight convolution network based on high‐altitude video for vehicle instantaneous counting in dense traffic scenes

“…In this section, we measure the inference time and FPS of GhostCount and other advanced networks on three platforms (RTX 3090‐GPU, R7 4800H‐CPU, and Raspberry Pi‐4B). The inference speed and FPS of the networks are measured according to the method [36]. The program runs 10 times.…”

“…Deep neural networks (DNNs) can fully extract the texture features of objects and can alleviate the counting problem under occlusion. Compared with various networks with complex structures, the work [36] achieved state‐of‐the‐art (SOTA) performance on the SHHB [9] dataset by simply removing the fully connected layer from ResNet‐101 and adding a decoder consisting of two superficial convolutional layers.…”

Analysis of the Effect of Low-Overhead Lossy Image Compression on the Performance of Visual Crowd Counting for Smart City Applications

“…The goal of crowd counting is to count the total number of people present in a given input image [1]. The input to crowd counting models is an image or a video frame, and the output is a density map showing the crowd density at each location of the image.…”

“…Crowd counting has many applications in video surveillance, social safety and crowd analysis and is an active area of research in the literature [1]. Since most crowd counting applications and datasets use surveillance footage, the input to crowd counting models are high-resolution images, typically Full HD (1,920×1,080 pixels) or even higher.…”

“…GigaZoom works by iteratively zooming into the densest areas of the image and refining the coarser density map with finer details. Our code is publicly available 1 .…”