A Comprehensive Survey of Weed Detection Methodologies in Soybean Crop

Weed detection is a significant problem, especially in India, where agriculture is one of the primary economic activities. Weeds are unwanted plants that compete with crops for resources, and act as pests to crops, resulting in reduced crop yield quality and increased cultivation and labor cost. So, effective site-specific weed detection and removal is necessary for efficient and cost-effective agriculture. Recent studies have focused on developing automated weed detection systems using machine learning, deep learning, and computer vision techniques. Most studies have focused on weed classification, with less attention given to identifying numerous weeds in a single image. The paper aims to comprehensively review and compare various weed detection techniques, focusing on soybean crops, as an image classification and object detection task. After examining the strengths and limitations of each approach, the study concludes that Convolutional Neural Network (CNN) based and Transformer-based deep learning models achieved the best possible accuracy possible. These methods can present a potential use case in precision agriculture.