| tiny object detection survey |
| Paper ID : 1108-ISCHU |
| Authors |
|
aya mohamed abdelrahman *1, MOURAD REEFAT2, Ahmeda Labib3, mona solyman4, ASHRAF DARWISH5 1Department of Mathematics, Faculty of Science, Helwan University, cairo, Egypt 2Faculty of scince,Helwan university, Cairo,Egypt 3faculty of scince,Helwan university,Cairo,Egypt 4Faculty of Computer and Information,Helwan university Cairo,Egypt 5Faculty of scince, Helwan university, Cairo,Egypt |
| Abstract |
| Tiny Object Detection (TOD) stands out as one of the most fundamental and formidable challenges within the realm of computer vision. In recent years, it has garnered significant attention, reflecting the swift technological evolution in the field of object detection and its profound impact on the broader landscape of computer vision. If we were to characterize today's object detection techniques, it would be as a revolution powered by deep learning. While several surveys have explored the domain of tiny object detection, most of them tend to be narrowly focused on specific application areas or specific classification tasks. What sets this survey apart is its broader perspective, seeking to appreciate the innovative thinking and long-term design philosophies that originated in the early days of computer vision. The primary objective of this survey is to comprehensively review the existing literature on tiny object detection, with a particular emphasis on the mathematical methods employed, which encompass probabilistic techniques, artificial intelligence approaches, and the theory of belief. The survey delves into the opportunities and challenges inherent in each of these mathematical methods and the environments in which they are applied. Furthermore, this survey provides an extensive examination of this rapidly evolving research field in light of its technical evolution. It covers a wide range of topics, including historical milestones in object detectors, the datasets used for detection, evaluation metrics, the fundamental components of detection systems, techniques for accelerating detection processes, and the most recent state-of-the-art detection methods. Additionally, the survey looks ahead to future developments and emerging areas that stand to gain significantly from advancements in tiny object detection. These include domains such as Traffic Video Surveillance, Small Object Retrieval, Anomaly Detection, Maritime Surveillance, Drone Surveying, Traffic Flow Analysis, and Object Tracking. The survey serves as a guidepost, providing insights into the current state of the field while also pointing towards promising avenues for future research and applications. |
| Keywords |
| Artificial intelligence-deep learning-computer vision -object detection - tiny object |
| Status: Abstract Accepted (Poster Presentation) |