Visual-infrared image monitoring system for moving object detection, classification and tracking using deep learning technique
Maad Issa Al Tameemi, Bashra Kadhim Oleiwi · AIP conference proceedings · 2023
In this work, the issue of moving object detection, classification, and tracking in both visual-infrared images and video streams is considered. The proposed system utilizing deep learning-based You Look Only Once -version3 (YOLOv3) algorithm as multilayers of CNN model, OpenCV, and CAMEL dataset and collected dataset as video sequence of a visual-infrared dataset. The proposed object detector system pre-trained on the dataset for streaming videos and different sizes and types of images as visual_ thermal as RGB images of different object classes and carried out by Python codes. The first part of the proposed system is to make the object or person detection and surround it by rectangle box, the center of the detected person is then specified in the second part, finally, calculate the relative accuracy of detected person and added it to the rectangle box. The obtained results proved the success of the proposed system in achieving objects detection, classification, and tracking within visual-infrared surveillance scenarios and gave satisfactory results with high accuracy.