Weed Net: Deep Learning Informed Convolutional Neural Network Based Weed Detection in Soybean Crops

This research employs Convolutional Neural Networks (CNNs) to distinguish grass and broadleaf weeds in soybean crops, addressing the challenge of weed interference in agriculture. A dataset of 15,336 curated images, featuring soil weeds, grass, broadleaf, and soybeans, forms a comprehensive image library. Utilizing deep learning architectures (ResNet, VGG16, VGG19, AlexNet, GoogleNet, LeNet), the trained neural networks achieved remarkable accuracies, with some models exceeding 97%, demonstrating robust weed identification for precision farming. The study's accuracy in detecting weeds, ranging from 90% to over 98%, highlights practical applicability in real-world scenarios. This research aids farmers in targeted weed management, promotes sustainable farming, and enhances crop yield. The paper details experimental design, evaluation metrics, dataset characteristics, and conclusive insights, offering a comprehensive overview of model effectiveness and potential impact on modern farming.