Disease Recognition in Plant Leaves Using CNN-Based Algorithm

• Published in: 2023 International Conference on Self Sustainable Artificial Intelligence Systems (ICSSAS) pp. 203 - 209
• Main Authors: Voditel, Preeti, Gurjar, Aparna, Kadoo, Pranjali, Dahake, Tanaya
• Format: Conference Proceeding
• Language: English
• Published: IEEE 18.10.2023
• Subjects: accuracy; classification; Classification algorithms; Convolution Neural Network; Convolutional neural networks; Crops; extraction; Feature extraction; plant disease; Plant diseases; quality; Transfer learning; Visualization
• DOI: 10.1109/ICSSAS57918.2023.10331750

In the agricultural sector, the scourge of plant diseases casts a long shadow, leading to substantial financial losses and threatening both crop quality and quantity. With global food demands on the rise and the expansion of crop cultivation to meet these needs, the urgency for effective, scalable, and prompt plant disease detection methods cannot be overstated. Traditional approaches reliant on manual visual inspections are fraught with labor intensiveness, time constraints, and inherent limitations, necessitating innovative solutions harnessing the capabilities of modern technology. The central goal of this system is to introduce a pragmatic, efficient, and adaptable approach that capitalizes on recent breakthroughs in neural network-based image analysis, with a specific focus on the transformative potential of Convolutional Neural Networks (CNNs). The proposed system encompasses several critical phases, commencing with image acquisition, followed by meticulous pre-processing, feature extraction, and culminating in classification via neural networks. What sets this system apart is its versatility, as it can be trained with a spectrum of algorithms, offering flexibility in adapting to various agricultural settings and unique plant disease scenarios. The outcomes resulting from the deployment of this multi-step plant disease detection system hold the promise of reshaping agriculture. It delivers timely and dependable insights into crop health, empowering farmers to make well-informed decisions and enabling the precision implementation of disease control strategies. By optimizing crop management practices through early disease identification, this system has the potential not only to mitigate the financial repercussions of diseases but also to bolster overall agricultural productivity. However, it is imperative to acknowledge that further research is indispensable. Such research should address potential limitations, fine-tune the system, and enhance its accuracy and effectiveness in diverse agricultural landscapes, thereby ensuring its pivotal role in shaping the future of sustainable and efficient crop management.