杂草
农学
杂草防治
作物产量
数学
作物
环境科学
生物
农业
农林复合经营
作者
Abhishek Rana,Vikas Attri,Punam Rattan
出处
期刊:
[Institute of Electrical and Electronics Engineers]
日期:2026-01-27
卷期号:: 1214-1221
被引量:2
标识
DOI:10.1109/icipcn67432.2026.11438422
摘要
Weed identification and classification are critical to enhance crop yield and aiding precision farming. The traditional manual methods of identifying weeds are labour-intensive and inefficient and therefore necessitate automated vision-based systems. This paper presents a deep learning-based classification system, which will help distinguish between rice crops and weeds. The model makes use of a field-gathered dataset of custom field dataset using a mobile camera in the natural environment conditions at Lovely Professional University as well some sampling images from Rice Field weed BD Dataset. Three convolutional models, namely CNN, VGG-16 and ResNet-18, were trained on an 80:20 dataset division using Stochastic Gradient Descent (SGD) and the CrossEntropyLoss function. The performance of the models was assessed using precision, recall, F1-score, loss convergence during training, and confusion matrices, which help justify the quality and generalization ability of the deeper models. The researchers assess the efficiency of the developed CNN models in detecting the presence of weeds in a farm's practice environment and emphasize the potential application of these neural networks in future automated precision farming programs.
科研通智能强力驱动
Strongly Powered by AbleSci AI