2025, Vol. 6, Issue 2, Part C

Crop health monitoring remains a major challenge in agriculture due to region-specific variations in diseases, insect infestations, and nutrient deficiencies. This paper proposes a deep learning-driven framework that provides precise crop intelligence across six major crops: Wheat, Paddy, Potato, Cotton, Mustard, and Kinnow. Transfer learning was applied on three pre-trained convolutional neural network models: ResNet-50, VGG16, and EfficientNet-B0, revealing that ResNet-50 achieved the highest accuracy of 95.2%. The model clearly distinguishes between diseases, insects, and nutrient deficiencies across different climates, with region-based data augmentation improving its flexibility. This is supported by experiments showing that the framework is scalable and capable of real-time performance, achieving inference times below 0.5 seconds per image. The proposed approach supports the way for precision agriculture by enabling early detection, targeted management, and scalable deployment across diverse agro-ecological zones.