Abstract:

Weeds are unwanted plants that grow among crops, and they can significantly harm farm output by lowering yield and quality. Unfortunately, site-specific weed management is not often used in farming practises. The authors of this paper investigate the classification of rice crops and weeds from digital photographs using several classifier systems, particularly support vector machines and random forest classifiers. Using this method, weeds can be identified and categorised, and suitable pesticides can then be suggested to farmers. For computer vision techniques to be successful, crops and weeds must be accurately identified and classified from digital images. Three cameras were used in rice fields by the authors to capture digital photos of weeds and paddy crops from various heights.After removing the backdrop from the digital photos, they extracted texture, colour, and shape attributes that were used for classification. The accuracy of the multiple classifier systems was higher than that of the single classifier systems because they used a straightforward and novel decision function

Description:

crops, yield reduction, support vector machines, random forest classifiers, computer vision.