Computational Efficiency in Precision Agriculture: MobileNetV2 Outperforms State-of-the-Art CNNs for Real-Time Plant Disease Detection

This research addresses critical challenges in global agriculture–particularly climate change, soil degradation, and water scarcity–by advancing smart farming technologies. Automated systems powered by computer vision (CV) and deep learning (DL) enable early plant disease detection, delivering faster, more accurate, and more cost-effective solutions than traditional methods. These innovations enhance crop management and food security. We present a comparative analysis of five deep learning architectures (EfficientNetV2, ResNet50, MobileNetV2, Inception-v2, and DenseNet210) for early pathology detection. By employing user-friendly imaging techniques and scalable solutions, this work facilitates widespread farmer adoption. Future efforts will refine these technologies, improve their adaptability across diverse agricultural environments, and integrate them into existing practices to boost productivity, sustainability, and resilience in modern agriculture.