scholarly journals Design of Efficient Human Head Statistics System in the Large-Angle Overlooking Scene

Electronics ◽  
2021 ◽  
Vol 10 (15) ◽  
pp. 1851
Author(s):  
An Wang ◽  
Xiaohong Cao ◽  
Lei Lu ◽  
Xinjing Zhou ◽  
Xuecheng Sun
Keyword(s):  
Smart Cities ◽  
Large Angle ◽  
Market Value ◽  
Human Head ◽  
Recall Rate ◽  
Frame Rate ◽  
Head Tracking ◽  
Scene Segmentation ◽  
Hash Algorithm ◽  
Head Detection

Human head statistics is widely used in the construction of smart cities and has great market value. In order to solve the problem of missing pedestrian features and poor statistics results in a large-angle overlooking scene, in this paper we propose a human head statistics system that consists of head detection, head tracking and head counting, where the proposed You-Only-Look-Once-Head (YOLOv5-H) network, improved from YOLOv5, is taken as the head detection benchmark, the DeepSORT algorithm with the Fusion-Hash algorithm for feature extraction (DeepSORT-FH) is proposed to track heads, and heads are counted by the proposed cross-boundary counting algorithm based on scene segmentation. Specifically, Complete-Intersection-over-Union (CIoU) is taken as the loss function of YOLOv5-H to make the predicted boxes more in line with the real boxes. The results demonstrate that the recall rate and [email protected] of the proposed YOLOv5-H can reach up to 94.3% and 93.1%, respectively, on the SCUT_HEAD dataset. The statistics system has an extremely low error rate of 3.5% on the TownCentreXVID dataset while maintaining a frame rate of 18FPS, which can meet the needs of human head statistics in monitoring scenarios and has a good application prospect.

Head Detection by the Adaboost Algorithm for Head Pose Estimation

2013 ◽  
Vol 427-429 ◽  
pp. 1696-1699
Author(s):  
Xiang Yang Liu ◽  
Shao Song Zhu ◽  
Su Qing Wu ◽  
Zhi Wei Shen
Keyword(s):  
Pose Estimation ◽  
Nearest Neighbor ◽  
Human Head ◽  
Head Pose Estimation ◽  
Head Tracking ◽  
Head Pose ◽  
Adaboost Algorithm ◽  
Head Detection ◽  
Dimensionality Reduction Method ◽  
Tracking Module

For human head pose analysis based on videos, pose is usually estimated on the head patch provided by a tracking module. However, head tracking is very sensitive to the large changes of pose. Therefore, this work locates the head patch in the videos by head detection. Firstly, we use the Adaboost algorithm to detect the human head in the video. Secondly, we present a dimensionality reduction method to process the head patch. Finally, we use the nearest neighbor method to estimate the head pose. The experiment results show: accurate head detecting helps to estimate the head pose. This method can be used for complex conditions of accurate head pose estimation.

scholarly journals A Systematic Deep Learning Based Overhead Tracking and Counting System Using RGB-D Remote Cameras

2021 ◽  
Vol 11 (12) ◽  
pp. 5503
Author(s):  
Munkhjargal Gochoo ◽  
Syeda Amna Rizwan ◽  
Yazeed Yasin Ghadi ◽  
Ahmad Jalal ◽  
Kibum Kim
Keyword(s):  
Deep Learning ◽  
Human Head ◽  
Point Clouds ◽  
Head Tracking ◽  
Complex Environments ◽  
People Tracking ◽  
Practical Applications ◽  
Remote Cameras ◽  
Video Frames ◽  
Benchmark Datasets

Automatic head tracking and counting using depth imagery has various practical applications in security, logistics, queue management, space utilization and visitor counting. However, no currently available system can clearly distinguish between a human head and other objects in order to track and count people accurately. For this reason, we propose a novel system that can track people by monitoring their heads and shoulders in complex environments and also count the number of people entering and exiting the scene. Our system is split into six phases; at first, preprocessing is done by converting videos of a scene into frames and removing the background from the video frames. Second, heads are detected using Hough Circular Gradient Transform, and shoulders are detected by HOG based symmetry methods. Third, three robust features, namely, fused joint HOG-LBP, Energy based Point clouds and Fused intra-inter trajectories are extracted. Fourth, the Apriori-Association is implemented to select the best features. Fifth, deep learning is used for accurate people tracking. Finally, heads are counted using Cross-line judgment. The system was tested on three benchmark datasets: the PCDS dataset, the MICC people counting dataset and the GOTPD dataset and counting accuracy of 98.40%, 98%, and 99% respectively was achieved. Our system obtained remarkable results.

scholarly journals Head Detection Based on DR Feature Extraction Network and Mixed Dilated Convolution Module

Electronics ◽  
2021 ◽  
Vol 10 (13) ◽  
pp. 1565
Author(s):  
Junwen Liu ◽  
Yongjun Zhang ◽  
Jianbin Xie ◽  
Yan Wei ◽  
Zewei Wang ◽  
...  
Keyword(s):  
Neural Network ◽  
Feature Extraction ◽  
Transmission Rate ◽  
Pedestrian Detection ◽  
Human Head ◽  
Detection Rates ◽  
Translational Invariance ◽  
Dilated Convolution ◽  
Head Detection ◽  
Small Targets

Pedestrian detection for complex scenes suffers from pedestrian occlusion issues, such as occlusions between pedestrians. As well-known, compared with the variability of the human body, the shape of a human head and their shoulders changes minimally and has high stability. Therefore, head detection is an important research area in the field of pedestrian detection. The translational invariance of neural network enables us to design a deep convolutional neural network, which means that, even if the appearance and location of the target changes, it can still be recognized effectively. However, the problems of scale invariance and high miss detection rates for small targets still exist. In this paper, a feature extraction network DR-Net based on Darknet-53 is proposed to improve the information transmission rate between convolutional layers and to extract more semantic information. In addition, the MDC (mixed dilated convolution) with different sampling rates of dilated convolution is embedded to improve the detection rate of small targets. We evaluated our method on three publicly available datasets and achieved excellent results. The AP (Average Precision) value on the Brainwash dataset, HollywoodHeads dataset, and SCUT-HEAD dataset reached 92.1%, 84.8%, and 90% respectively.

Modeling head tracking of visual targets

2002 ◽  
Vol 12 (1) ◽  
pp. 25-33
Author(s):  
K.J. Chen ◽  
E.A. Keshner ◽  
B.W. Peterson ◽  
T.C. Hain
Keyword(s):  
Control System ◽  
Visual Feedback ◽  
Human Head ◽  
Feedback System ◽  
Optimization Methods ◽  
Head Tracking ◽  
Feedback Systems ◽  
Control Gain ◽  
Head Control ◽  
Visual Targets

Control of the head involves somatosensory, vestibular, and visual feedback. The dynamics of these three feedback systems must be identified in order to gain a greater understanding of the head control system. We have completed one step in the development of a head control model by identifying the dynamics of the visual feedback system. A mathematical model of human head tracking of visual targets in the horizontal plane was fit to experimental data from seven subjects performing a visual head tracking task. The model incorporates components based on the underlying physiology of the head control system. Using optimization methods, we were able to identify neural processing delay, visual control gain, and neck viscosity parameters in each experimental subject.

Mechanisms controlling human head stabilization. I. Head-neck dynamics during random rotations in the horizontal plane

1995 ◽  
Vol 73 (6) ◽  
pp. 2293-2301 ◽  
Author(s):  
F. A. Keshner ◽  
B. W. Peterson
Keyword(s):  
Visual Feedback ◽  
Horizontal Plane ◽  
Mental Arithmetic ◽  
Human Head ◽  
The Body ◽  
Myoelectric Activity ◽  
Head Tracking ◽  
Arithmetic Task ◽  
Head Velocity ◽  
Head Neck

1. Potential mechanisms for controlling stabilization of the head and neck include voluntary movements, vestibular (VCR) and proprioceptive (CCR) neck reflexes, and system mechanics. In this study we have tested the hypothesis that the relative importance of those mechanisms in producing compensatory actions of the head-neck motor system depends on the frequency of an externally applied perturbation. Angular velocity of the head with respect to the trunk (neck) and myoelectric activity of three neck muscles were recorded in seven seated subjects during pseudorandom rotations of the trunk in the horizontal plane. Subjects were externally perturbed with a random sum-of-sines stimulus at frequencies ranging from 0.185 to 4.11 Hz. Four instructional sets were presented. Voluntary mechanisms were examined by having the subjects actively stabilize the head in the presence of visual feedback as the body was rotated (VS). Visual feedback was then removed, and the subjects attempted to stabilize the head in the dark as the body was rotated (NV). Reflex mechanisms were examined when subjects performed a mental arithmetic task during body rotations in the dark (MA). Finally, subjects performed a voluntary head tracking task while the body was kept stationary (VT). 2. Gains and phases of head velocity indicated good compensation to the stimulus in VS and NV at frequencies < 1 Hz. Gains dropped and phases advanced between 1 and 2 Hz, suggesting interference between neural and mechanical components. Above 3 Hz, the gains of head velocity increased steeply and exceeded unity, suggesting the emergence of mechanical resonance.(ABSTRACT TRUNCATED AT 250 WORDS)

Human head detection using multi-modal object features

2004 ◽  
Author(s):  
Yun Luo ◽  
Yi Lu Murphey ◽  
F. Khairallah
Keyword(s):  
Human Head ◽  
Head Detection ◽  
Object Features

scholarly journals Challenges and Limitations in Human Action Recognition on Unmanned Aerial Vehicles: A Comprehensive Survey

2021 ◽  
Vol 38 (5) ◽  
pp. 1403-1411
Author(s):  
Nashwan Adnan Othman ◽  
Ilhan Aydin
Keyword(s):  
Action Recognition ◽  
Smart Cities ◽  
Large Angle ◽  
Human Action Recognition ◽  
Human Action ◽  
Human Detection ◽  
Long Distance ◽  
Benchmark Datasets ◽  
Comprehensive Survey ◽  
Aerial Vehicle

An Unmanned Aerial Vehicle (UAV), commonly called a drone, is an aircraft without a human pilot aboard. Making UAVs that can accurately discover individuals on the ground is very important for various applications, such as people searches, and surveillance. UAV integration in smart cities is challenging, however, because of problems and concerns such as privacy, safety, and ethical/legal use. Human action recognition-based UAVs can utilize modern technologies. Thus, it is essential for future development of the aforementioned applications. UAV-based human activity recognition is the procedure of classifying photo sequences with action labels. This paper offers a comprehensive study of UAV-based human action recognition techniques. Furthermore, we conduct empirical research studies to assess several factors that might influence the efficiency of human detection and action recognition techniques in UAVs. Benchmark datasets commonly utilized for UAV-based human action recognition are briefly explained. Our findings reveal that the existing human action recognition innovations can identify human actions on UAVs with some limitations in range, altitudes, long-distance, and a large angle of depression.

Sign in / Sign up

Export Citation Format

Share Document