Real Time Video Object Detection Using Deep Learning

• Published in: 2023 International Conference on Advances in Computation, Communication and Information Technology (ICAICCIT) pp. 601 - 606
• Main Authors: Sharma, Kunal, Sahu, Tanmay Kumar, Bhatt, Prateek, Singh, Aniket Kumar, Bhatia, Madhulika
• Format: Conference Proceeding
• Language: English
• Published: IEEE 23.11.2023
• Subjects: Analytical models; Deep learning; Learning; Neural networks; Object detection; pipeline processing; Reviews; Streaming media; Transforms; video signal processing; YOLO
• DOI: 10.1109/ICAICCIT60255.2023.10466207

The review of "Real-Time Video Object Detection using Deep Learning" provides an extensive analysis of the state-of-the-art in deep learning-powered real-time video object recognition systems. It examines the development of object identification models, stressing significant improvements in model accuracy, efficiency, and architecture design. It thoroughly investigates the most recent approaches and strategies for real-time video object detection. YOLO, Faster R-CNN, and SSD are just a few of the well-known deep learning models that are examined in this article to highlight their advantages and disadvantages in the context of real-time video analysis. It also explores the fundamental issues and most recent developments that have enabled deep learning to completely transform real-time object detection. Furthermore, this review broadens its scope to include additional noteworthy designs, such MobileNetV2 and Efficient D4, in recognition of the increasing demand for precise and effective real-time video analysis across applications like augmented reality, autonomous driving, and surveillance. This review provides a thorough evaluation of these models' capabilities by looking at their underlying theories and technological foundations, evaluating how well they perform on benchmark datasets, and considering metrics like speed, accuracy, and memory efficiency. It is noteworthy that this study employs and compares current models for real-time video object detection, making it a valuable resource for researchers, practitioners, and enthusiasts alike. This comprehensive analysis offers guidance for upcoming advancements in this quickly developing subject in addition to insights into the state-of-the-art models.