Image recognition using artificial intelligence(multi-object tracking using artificial intelligence)

• Main Author: Su, Hang
• Other Authors: Yap Kim Hui
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/167840
• Institution: Nanyang Technological University

With the continuous advancement of modern scientific and technological developments, life services, industrial production and safety monitoring are moving towards an era of unmanned and intelligent operations. Therefore, to achieve such applications, monitoring and tracking of human activities is of vital importance. Multiple Object Tracking, as one of the key technologies in computer vision, can be used in various applications. This final year project aims to design an effective multiple object tracking model for video sequences so that it may be applied to different applications. The project used a dataset of challenging real-world scenarios to evaluate the performance of the model and the evaluating benchmark is MOTChallenge. Performance of state-of-the-art object detection and multiple object tracking models are investigated by a thorough survey. The baseline tracking algorithm selected was ByteTrack and BoT-SORT, which employs deep learning techniques to track objects over time by associating object detection bounding boxes across frames. After selecting the model, different implementation settings and model backbones are being explored in detail to increase the model performance. The results shows that the proposed model accuracy increased from 74.08% and achieved 77.02% accuracy. The system can be used in a variety of applications, including traffic monitoring, game players monitoring and surveillance and so on .