An Assessment of Object Detection in Thermal (Infrared) Image Processing

Computer vision is used to detect objects in clumps of space. This paper explores the potential of automatically identifying people and their actions in both moving and static images. Then describe a number of issues with human identification and activity recognition in terms of various processing, segmentation, and extraction procedures for human attributes. Additionally, examine the issues with each phase and present the most up-to-date techniques, gaps between them, technological obstacles, applications, and their limitations. Object motion activities are not accurately detected at night, for object detection and classification, shape-based detection, skin color extraction, edge detection, and an object detector are used. Tracking objects using thermal images has been used in a variety of settings, including surveillance systems, spotting pedestrians, spotting wildlife, and more. An evaluation of the most current developments in object detectors like YOLOV7 is planned. This would result in increased detection accuracy and quicker inference speed. To analyze the various approaches to object detection, a future research direction for detecting objects (humans) in thermal video or images was proposed.