| Sumario: | India loses 35% of the yearly crop productivity owing to plant diseases. Earlier plant disease detection using traditional methods or human experts is a complex and time-consuming process. Therefore, rapid and automated plant disease detection models are essential to meet the increasing demand for food productivity and quality. Presently, computer vision and image processing techniques find useful for plant disease detection and increase crop yield sustainably. Therefore, this paper attempts to propose an efficient handcrafted feature with machine learning based plant leaf disease diagnosis and classification model. The proposed model uses a Gaussian filtering (GF) technique to preprocess the input image and boosts its quality. Besides, Grabcut based segmentation technique is utilized to identify the diseased portions in the plant leaves. Moreover, two feature extractors namely local binary patterns (LBP) and Scale Invariant Feature Transform (SIFT) models are applied as feature extractors. At last, multilayer perceptron (MLP) and random forest (RF) models are employed as the classifier models to allocate the proper class labels to the test plant leaf images. The performance of proposed method is assessed against a benchmark plant leaf disease dataset and the experimental outcomes show the promising efficiency of the proposed model over the recent methods interms of different measures.
|
|---|