Parallel deep CNN for tomato leaf disease detection

Agriculture faces significant risks from plant diseases and venomous insects, highlighting the crucial need for swift detection and diagnosis of these disorders. Continuous advancements in deep learning (DL) techniques have significantly facilitated the identification of plant leaf diseases, providing accurate and powerful tools. The accuracy of DL methods heavily depends on the quality and quantity of labeled samples used during training. This article introduces Tomato leaf disease detection using a Parallel Deep Convolutional Neural Network (TPDCNN) for plant leaf disease detection (PLDD). Additionally, it presents the use of a Conditional Generative Adversarial Neural Network (C-GAN) for generating artificial data to address the issue of limited data availability caused by imbalanced dataset sizes. Experimental results are conducted using the PlantVillage dataset (tomato plants), focusing on two-, six-, and ten-class PLDD. The effectiveness of the TPDCNN model is evaluated through various performance measures, including accuracy, recall, precision, and F1-score, and compared against traditional state-of-the-art approaches used for detecting tomato plant leaf diseases. The proposed system outperforms existing methods, achieving superior accuracy rates (99.14% for 2-class, 99.05% for 6-class, 98.11% for 10-class PLDD) for tomato PLDD. The TPDCNN method is well-suited for real-time deployment on standalone devices with limited computational resources due to its simpler structure and fewer trainable parameters.