Smart Glass: Real-Time Leaf Disease Detection using YOLO Transfer Learning
Vijayakumar Ponnusamy, Amrith Coumaran, Akhash Subramanian Shunmugam, K.V. Rajaram, Sanoj Senthilvelavan
Abstract
Having a keen observation and recognizing patterns in minute things is an arduous task in today's fast-paced world. These patterns might contain information that might be of significance to humans. In order to harness these regularities and to predict the activities in the near future, there exists numerous Artificial Neural Network architectures that have high degrees of accuracy. The drawback? These architectures demand GPUs with high processing capabilities, which in turn increase the size of the overall system. Existing systems capable of processing Machine Learning algorithms are not economical, or are not portable. Combining the merits of mobility, and a state-of-the-art You Only Look Once(YOLOv3) object detection system, we present to you, a Smart Glass, capable of highly accurate binary classification of data in real-time. Training the architecture with agricultural data, this wearable device would be able to identify Healthy and Unhealthy plant leaves in Real-time. Researchers could train the architecture with different datasets to obtain solutions to a wide variety of problems in the Agriculture, Healthcare, Automobile Industry, etc.