Hyderabad: Having seen first-hand the devastating impact of plant diseases on crops like tomato while growing up in a farmer’s family, a computer science faculty member from Hyderabad secured a patent for an AI-driven innovation, “Leaf Disease Detection System Using Convolutional Neural Networks.”The patented system, which was the result of collaborative efforts by researchers from the departments of Computer Science & Engineering, Information Technology, Electronics & Electrical Engineering, and Electronics & Communication Engineering at VNR Vignana Jyothi Institute of Engineering & Technology (VNRVJIET), Hyderabad, can detect leaf diseases at an early stage, enabling timely intervention and reducing crop losses.“When I was about 13, my father cultivated tomato crop in three out of five acres of our land. I clearly remember that we lost the entire crop because there was no means for early detection of crop diseases. Even at that age, I thought of why such a solution could not be made available for farmers,” said Vijaya Saraswathi R, the lead researcher.Almost two decades later, after securing her PhD, Saraswathi, along with ten faculty members, started working on a solution using artificial intelligence, deep learning, and Convolutional Neural Networks to help farmers.“While similar solutions might be available, our methodology is different. And, for the same reason, we were awarded a patent. When it comes to image resizing, data augmentation, and noise reduction, among other aspects, we have used a unique methodology. The same goes for the techniques used for image acquisition,” she added.Though the team used its own data for testing the product, it used a dataset of about 20,000 images of tomato, potato, and pepper plants to train the model to identify the 10 most prevalent diseases in these plants. The system also suggests the type of pesticide to be used for a particular disease to help farmers make informed decisions, optimise pesticide usage, and enhance agricultural productivity.The team is now gearing up to bring the solution in the form of a mobile application so that it can be handy for farmers. The patent was granted about two months ago. According to the researchers, the model achieved 96% accuracy in detecting leaf diseases.
