Review of Weed Detection Methods Based on Computer Vision

Abstract

Weeds are one of the most important factors affecting agricultural production. The waste and pollution of farmland ecological environment caused by full-coverage chemical herbicide spraying are becoming increasingly evident. With the continuous improvement in the agricultural production level, accurately distinguishing crops from weeds and achieving precise spraying only for weeds are important. However, precise spraying depends on accurately identifying and locating weeds and crops. In recent years, some scholars have used various computer vision methods to achieve this purpose. This review elaborates the two aspects of using traditional image-processing methods and deep learning-based methods to solve weed detection problems. It provides an overview of various methods for weed detection in recent years, analyzes the advantages and disadvantages of existing methods, and introduces several related plant leaves, weed datasets, and weeding machinery. Lastly, the problems and difficulties of the existing weed detection methods are analyzed, and the development trend of future research is prospected.

Using Remote Sensing and an Unmanned Aerial System for Weed Management in Agricultural Crops: A Review

“…OBIA creates 'objects' by grouping adjacent pixels with homogeneous spectral values, then combines them spectrally, topologically, and contextually to improve the precision of the image classification [43]. According to Wu et al [44], the position of vegetation objects in the crop row structure is critical for accurate weed detection for herbicide prescription maps in wide row crops such as maize. Other barriers in early season weed mapping involve ultra-high spatial resolution imaging and the need for a timely post-emergency control in a short period of time.…”

The Predictive Power of Regression Models to Determine Grass Weed Infestations in Cereals Based on Drone Imagery—Statistical and Practical Aspects

“…Weed detection is an area that is in rapid development and that has been taking advantage of the developments in machine learning (ML) and deep learning (DL) algorithms [11,12]. The present work focused on a small area in weed detection: selecting a proper training set to optimize the prediction accuracy.…”

“…Both methods rely on accurate weed monitoring and a variety of sensing devices, such as optoelectronic sensors. The use of RGB multispectral and hyperspectral cameras for weed detection have been thoroughly investigated [10,11].…”

“…The literature is full of agricultural applications of drones [17] and weed detection based on computer vision [2,11,18], but this study does not aim to contribute with cuttingedge knowledge on drone technology or machine learning. This work is based on tried and tested drone RGB imagery and the well-established knowledge that clearly visible weeds with different colors compared to the background have distinct spectral signatures.…”

A Lightweight Deep Network for Defect Detection of Insert Molding Based on X-ray Imaging

“…Recently, machine vision has been popularly used in the defect detection of industrial products instead of labor. However, most studies are focused on the detection of products’ external surface [ 4 , 5 , 6 , 7 ], while few are reported on the internal defects of injection-molded parts with DR imaging. Early work in internal defect detection with DR images using machine vision was based on traditional image processing for automated supervision and localization of defects, which mainly relies on manually produced feature extractors, such as area feature extraction, edge detection, threshold segmentation [ 8 ].…”