A Hybrid Swarm Intelligence Convolution Neural Network for Object Detection and Tracking
No Thumbnail Available
Date
2023-12
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Lead City University
Abstract
In recent years, analysis and interpretation of video sequences to detect and track objects of interest had become an active research field in computer vision and image processing. Despite significant efforts in object detection and tracking, an efficient method which provides high computational efficiency has not been developed. Hence, in this work a Hybrid Particle Swarm Optimization Convolution Neural Network (CNN-HPSO) technique was developed to improve computational efficiency in object detection and tracking. The video datasets (MP4 and AVi video formats) used in this work were obtained from a conventional online database and on a real-time basis from YouTube. Multiple frames sampled from the video clips were pre-processed and
then segmented. An Enhanced Particle Swarm Optimization (HPSO) was formulated from standard PSO and was applied to Convolution Neural Network (CNN). CNN- HPSO technique was used for edge detection and extraction of the boundary of the image and the object tracking was finally carried out. The work was implemented using MatLab R2016 software. The average results of CNN-HPSO, CNN-PSO and CNN on the videos with MP4 format yielded processing time, accuracy, precision, FPR, sensitivity and specificity of 165.89s, 97.08%, 98.41%, 7.75%, 97.82%, and 92.25%; 179.52s, 94.25%, 96.99%, 10.37%, 95.23% and 89.62%; and 189.19s, 89.95%, 93.64%, 15.56%, 91.95.33% and 84.44% respectively. For the videos in AVi
format, CNN-HPSO, CNN-PSO and CNN produced similar average results with processing time, accuracy, precision, FPR, sensitivity, and specificity of 185.09s, 96.62%, 98.23%, 7.80%, 97.34% and 92.19%; 198.24s, 94.83%, 97.62%, 8.56%, 95.46% and 91.43%; and 216.59s, 91.30%, 93.98%, 15.09%, 93.67% and 84.91% respectively. In this research, a CNN-HPSO with associated high computational efficiency was developed. The developed technique can be used for solving other related optimization problems.
Keywords: Deep Learning Algorithms, Computer Vision, Moving Objects, Video Frame, Object Segmentation
Word Count: 293
Description
Keywords
Deep Learning Algorithms, Computer Vision, Moving Objects, Video Frame, Object Segmentation