AI-Driven-IoT (AIIoT)-Based Jawar Leaf Disease Detection
Kutubuddin Sayyad Liyakat Kazi
Abstract
This study evaluated the efficacy of three different models for identifying diseases that affect jawar leaves: Decision Trees (DT), K-Nearest Neighbours (K-NN), and Artificial Neural Networks (ANN). Both healthy and sick specimens of jawar leaves might be included in the dataset of jawar leaf photographs that could be obtained. A number of different image processing techniques were utilised in order to extract characteristic information from the photographs. Once the features had been extracted, the ANN, DT, and K-NN models were trained and evaluated using the information that was obtained. In addition to having the best accuracy (96.5%), the KSK technique had the highest recall (95.8%) and precision (97.2%) measurements. In comparison to the DT and K-NN models, the ANN model performed significantly better. This was due to the fact that it was able to analyse the data and identify subtle non-linear relationships.