Cotton Detection Using YOLOv5

Abstract

Cotton Harvesting has always been a labor-intensive and time-consuming task which involved significant challenges. Manual harvesting resulted in inconsistencies in yield and quality. In traditional systems the cotton blooms could not be detected accurately due to obstruction due to leaves, or detecting the sky instead of the cotton based on features like color, etc. To address this issue Cotton Harvesting Rover implements a robotic system which can detect cotton blooms accurately using computer vision technology. The system autonomously identifies the cotton blooms amongst the fields based on particular features and then picks them using a robotic arm. This study reviews the state-of-the-art deep learning methods for object detection. This review paper will provide a comprehensive study of various deep learning methods which can be useful for automatic detection and various image processing methodologies. Additionally, it offers performance indicators for every technique used, including F-1 score, accuracy, and precision. The climatic condition for cotton productivity in various regions has been discussed. This paper gives a review about the alternative to traditional methods which needed prior knowledge, processing of large data, and reducingdifficulties caused due to computer hardware.