scholarly journals Integer arithmetic approximation of the hog algorithm used for pedestrian detection

2017 ◽  
Vol 14 (2) ◽  
pp. 329-346 ◽  
Author(s):  
Srdjan Sladojevic ◽  
Andras Anderla ◽  
Dubravko Culibrk ◽  
Darko Stefanovic ◽  
Bojan Lalic
Keyword(s):  
Classification Accuracy ◽  
System Performance ◽  
Detection System ◽  
Pedestrian Detection ◽  
Lessons Learned ◽  
People Detection ◽  
Algorithm Performance ◽  
Integer Arithmetic ◽  
Point Arithmetic ◽  
Angle Calculation

This paper presents the results of a study of the effects of integer (fixed-point) arithmetic implementation on classification accuracy of a popular open-source people detection system based on Histogram of Oriented Gradients. It is investigated how the system performance deviates from the reference algorithm performance as integer arithmetic is introduced with different bit-width in several critical parts of the system. In performed experiments, the effects of different bit-width integer arithmetic implementation for four key operations were separately considered: HoG descriptor magnitude calculation, HoG descriptor angle calculation, normalization and SVM classification. It is found that a 13-bit representation of variables is more than sufficient to accurately implement this system in integer arithmetic. The experiments in the paper are conducted for pedestrian detection and the methodology and the lessons learned from this study allow generalization of conclusions to a broader class of applications.

The Impact of the Detector on the Performances of a Multi Person Tracking System

2021 ◽  
Vol 15 ◽  
Author(s):  
Djalal Djarah ◽  
Abdallah Meraoumia ◽  
Mohamed Lakhdar Louazene
Keyword(s):  
System Performance ◽  
Detection System ◽  
Tracking System ◽  
Pedestrian Detection ◽  
Dynamic Environment ◽  
Human Gait ◽  
Detection And Tracking ◽  
Tracking By Detection ◽  
Multi Person Tracking ◽  
The Impact

Background: Pedestrian detection and tracking is an important area of study in real-world applications such as mobile robots, human-computer interaction, video surveillance, pedestrian protection systems, etc. As a result, it has attracted the interest of the scientific community. Objective: Certainly, tracking people is critical for numerous utility areas which cover unusual situations detection, like vicinity evaluation and sometimes change direction in human gait and partial occlusions. Researchers primary focus is to develop surveillance system that can work in a dynamic environment, but there are major issues and challenges involved in designing such systems. So, it has become a major issue and challenge to design a tracking system that can be more suitable for such situations. To this end, this paper presents a comparative evaluation of the tracking-by-detection system along with the publicly available pedestrian benchmark databases. Method: Unlike recent works where the person detection and tracking are usually treated separately, our work explores the joint use of the popular Simple Online and Real-time Tracking (SORT) method and the relevant visual detectors. Consequently, the choice of the detector is an important factor in the evaluation of the system performance. Results: Experimental results demonstrate that the performance of the tracking-by-detection system is closely related to the optimal selection of the detector and should be required prior to a rigorous evaluation. Conclusion: The study demonstrates how sensitive the system performance as a whole is to the challenging of the dataset. Furthermore, the efficiency of the detector and the detector-tracker combination are also depending on the dataset.

scholarly journals Multi-Scale Feature Pyramid Network: A Heavily Occluded Pedestrian Detection Network Based on ResNet

Sensors ◽  
2021 ◽  
Vol 21 (5) ◽  
pp. 1820
Author(s):  
Xiaotao Shao ◽  
Qing Wang ◽  
Wei Yang ◽  
Yun Chen ◽  
Yi Xie ◽  
...  
Keyword(s):  
Semantic Information ◽  
Detection System ◽  
Pedestrian Detection ◽  
Detection Accuracy ◽  
The Public ◽  
Scale Feature ◽  
Detection Algorithms ◽  
Multi Scale ◽  
Art Works ◽  
Feature Pyramid

The existing pedestrian detection algorithms cannot effectively extract features of heavily occluded targets which results in lower detection accuracy. To solve the heavy occlusion in crowds, we propose a multi-scale feature pyramid network based on ResNet (MFPN) to enhance the features of occluded targets and improve the detection accuracy. MFPN includes two modules, namely double feature pyramid network (FPN) integrated with ResNet (DFR) and repulsion loss of minimum (RLM). We propose the double FPN which improves the architecture to further enhance the semantic information and contours of occluded pedestrians, and provide a new way for feature extraction of occluded targets. The features extracted by our network can be more separated and clearer, especially those heavily occluded pedestrians. Repulsion loss is introduced to improve the loss function which can keep predicted boxes away from the ground truths of the unrelated targets. Experiments carried out on the public CrowdHuman dataset, we obtain 90.96% AP which yields the best performance, 5.16% AP gains compared to the FPN-ResNet50 baseline. Compared with the state-of-the-art works, the performance of the pedestrian detection system has been boosted with our method.

scholarly journals Random Forest with Adaptive Local Template for Pedestrian Detection

2015 ◽  
Vol 2015 ◽  
pp. 1-11 ◽  
Author(s):  
Tao Xiang ◽  
Tao Li ◽  
Mao Ye ◽  
Zijian Liu
Keyword(s):  
Computer Vision ◽  
Random Forest ◽  
Classification Accuracy ◽  
Template Matching ◽  
Detection Method ◽  
State Of The Art ◽  
Pedestrian Detection ◽  
Sliding Window ◽  
Experimental Results ◽  
Training Samples

Pedestrian detection with large intraclass variations is still a challenging task in computer vision. In this paper, we propose a novel pedestrian detection method based on Random Forest. Firstly, we generate a few local templates with different sizes and different locations in positive exemplars. Then, the Random Forest is built whose splitting functions are optimized by maximizing class purity of matching the local templates to the training samples, respectively. To improve the classification accuracy, we adopt a boosting-like algorithm to update the weights of the training samples in a layer-wise fashion. During detection, the trained Random Forest will vote the category when a sliding window is input. Our contributions are the splitting functions based on local template matching with adaptive size and location and iteratively weight updating method. We evaluate the proposed method on 2 well-known challenging datasets: TUD pedestrians and INRIA pedestrians. The experimental results demonstrate that our method achieves state-of-the-art or competitive performance.

A new thermal infrared and visible spectrum images-based pedestrian detection system

2018 ◽  
Vol 78 (12) ◽  
pp. 15861-15885 ◽  
Author(s):  
Redouan Lahmyed ◽  
Mohamed El Ansari ◽  
Ayoub Ellahyani
Keyword(s):  
Detection System ◽  
Pedestrian Detection ◽  
Visible Spectrum ◽  
Thermal Infrared

scholarly journals An Indonesian Hoax News Detection System Using Reader Feedback and Naïve Bayes Algorithm

2020 ◽  
Vol 20 (1) ◽  
pp. 82-94
Author(s):  
Badrus Zaman ◽  
Army Justitia ◽  
Kretawiweka Nuraga Sani ◽  
Endah Purwanti
Keyword(s):  
Performance Evaluation ◽  
System Performance ◽  
Naive Bayes ◽  
Detection System ◽  
Naïve Bayes ◽  
Bayes Algorithm ◽  
F Measure ◽  
System Performance Evaluation

AbstractHoax news in Indonesia spread at an alarming rate. To reduce this, hoax news detection system needs to be created and put into practice. Such a system may use readers’ feedback and Naïve Bayes algorithm, which is used to verify news. Overtime, by using readers’ feedback, database corpus will continue to grow and could improve system performance. The current research aims to reach this. System performance evaluation is carried out under two conditions ‒ with and without sources (URL). The system is able to detect hoax news very well under both conditions. The highest precision, recall and f-measure values when including URL are 0.91, 1, and 0.95 respectively. Meanwhile, the highest value of precision, recall and f-measure without URL are 0.88, 1 and 0.94, respectively.

Sign in / Sign up

Export Citation Format

Share Document