ZoomCount: A Zooming Mechanism for Crowd Counting in Static Images

Abstract

This paper proposes a novel approach for crowd counting in low to high density scenarios in static images. Current approaches cannot handle huge crowd diversity well and thus perform poorly in extreme cases, where the crowd density in different regions of an image is either too low or too high, leading to crowd underestimation or overestimation. The proposed solution is based on the observation that detecting and handling such extreme cases in a specialized way leads to better crowd estimation. Additionally, existing methods find it hard to differentiate between the actual crowd and the cluttered background regions, resulting in further count overestimation. To address these issues, we propose a simple yet effective modular approach, where an input image is first subdivided into fixedsize patches and then fed to a four-way classification module labeling each image patch as low, medium, high-dense or nocrowd. This module also provides a count for each label, which is then analyzed via a specifically devised novel decision module to decide whether the image belongs to any of the two extreme cases (very low or very high density) or a normal case. Images, specified as high-or low-density extreme or a normal case, pass through dedicated zooming or normal patch-making blocks respectively before routing to the regressor in the form of fixed-size patches for crowd estimate. Extensive experimental evaluations demonstrate that the proposed approach outperforms the state-of-the-art methods on four benchmarks under most of the evaluation criteria.

S$^2$FPR: Crowd Counting via Self-Supervised Coarse to Fine Feature Pyramid Ranking

“…Bai et al [52] self-corrected the density map by EM algorithm. ZoomCount [28] proposed a zooming mechanism to tackle the underestimation and overestimation issues due to the density variation problem. Adversarial networks [53], [54], [55] are also used for crowd counting to generate a high-quality density map.…”

“…Benefiting from the strong representation learning ability of convolutional neural networks (CNN) [10], [11], [12], [13], [14], CNN-based methods [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31] are employed to predict a density map of a still image because the density map contains more spatial information of people distribution and its integral equals the number of people in one image. For example, multi-branch architectures [16], [18], [17], [19] are designed to extract the multi-scale features and detect varying sizes of heads because different-sized convolutional filters have varying receptive fields, which are more useful for learning non-uniform crowd distribution.…”

A comparative study on polyp classification using convolutional neural networks

“…Deep learning models : Inspired by the success of AlexNet [ 16 ] in the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) 2012, convolutional neural networks (CNN) have attracted a lot of attention and been successfully applied to image classification [ 20 – 22 ], object detection [ 4 , 23 , 24 ], depth estimation [ 25 , 26 ], image transformation [ 27 , 28 ], and crowd counting [ 29 ]citesajid2020plug. VGGNets [ 14 ], and GoogleNet [ 17 ], the ILSVRC winners of 2014 and 2015, proved that deeper models could significantly increase the ability of representations.…”

Location-Aware Box Reasoning for Anchor-Based Single-Shot Object Detection

“…D EEP networks have been dramatically driving the progress of computer vision, bringing out a series of popular models for different vision tasks [35] [31], like image classification [3] [29], object detection [32] [15], crowd counting [25], depth estimation [10], and image translation [30]. Object detection plays an important role and serves as a prerequisite for numerous computer vision applications, such as instance segmentation, face recognition, autonomous driving, and video analysis [1], [9], [11], [12], [21].…”