Camouflaged Human Detection: Comparative Analysis Using YoloV5s, Yolov5l, Yolov5m, Yolov5n, Yolov5x Model

This research paper discusses about, The Camouflaged human or object Detection using different variations of YOLOV5. Our project deep dives into the analysis of YOLO(You Only Look Once) algorithm and it's different types for performance analysis. The proposed model, trained on the dataset with different regions and angles of images, utilizes YOLOv5 for efficient camouflaged human detection, with manual annotation of images used for training the dataset and calculated different outputs for performance analysis. There can be lot of applications of the Camouflaged object detection project. Even if it is Biodiversity conservation in Nature like detecting the animals or plants which gets camouflaged in the nature and are difficult to detect or detecting enemy military invasion. This algorithm can be combined with one or more other image processing algorithms also for better accuracy and easy implementation purpose. This research work is carried out on different variants of YOLOv5 algorithm. We have used the different variants like Yolov5l, Yolov5m, Yolov5n, Yolov5x enhancing accuracy and robustness. The performance analysis is carried out in terms of P curve (Precision), R curve(recall), PR curve(precision and recall) and F1 curve of each training set and also the confusion matrix for all the algorithms is calculated and it is observed that Yolov5l works better as compared to other YOLO variants.