Implementation of YOLOv5 Architecture for Clothing Detection Systems

Abstract

Object detection is one of the key areas in computer vision that plays a crucial role in processing and analyzing visual data. In various applications such as clothing recognition, object detection is instrumental in identifying and localizing objects in image or videos. This research untilizes one of the Convolutional Neural Network (CNN) architectures, YOLOv5n, to debelop an effective framework for clothing detection. The objective is to enchance the performance of YOLOv5n in terms of accuracy while ensuring applicability to low-cost devices. Additionally, a new clothing dataset is curated for this purpose. The study leverages CPU-based cameras for real-time object detection. By modifying the SPPF module within the YOLOv5n architecture, high accuracy is achieved with parameters totaling 7,086,779, mAP of 0,685, and GFLOPS of 8,3.