Abstract

BHUVANA, J.  and  MIRNALINEE, T.T.. An approach to Plant Disease Detection using Deep Learning Techniques. Iteckne [online]. 2021, vol.18, n.2, pp.161-169.  Epub May 16, 2022. ISSN 1692-1798.  https://doi.org/10.15332/iteckne.v18i2.2615.

Agriculture is the backbone of Indian economy. Conventional farming systems are no longer being followed by our generation, due to lack of knowledge and expertise. Advancement of technologies pave a path that make a transition from traditional farming methods to smart agriculture by automating the processes involved. Challenges faced by today's agriculture are depletion of soil nutrients and diseases caused by pests which lead to low productivity, irrigation problems, soil erosion, shortage of storage facilities, availability of quality seeds, lack of transportation, poor marketing etc. Among all these challenges in agriculture, prediction of diseases remains a major issue to be addressed. Identifying diseases based on visual inspection is the traditional way of farming which needs knowledge and experience to handle. Automating the process of detecting and identifying through visual inspection (cognitive) is the motivation behind this work. This is made possible with the availability of images of the plant or parts of plants, since most diseases are reflected on the leaves. A deep learning network architecture named Plant Disease Detection Network PDDNet-cv and a transfer learning approach of identifying diseases in plants were proposed. Our proposed system is compared with VGG19, ResNet50, InceptionResNetV2, the state-of-the-art methods reported in [9, 13, 5] and the results show that our method is significantly performing better than the existing systems. Our proposed PDDNet-cv has achieved average classification accuracy of 99.09% in detecting different classes of diseases. The proposed not so deep architecture is performing well compared to other deep learning architectures in terms of performance and computational time.

Keywords : Plant disease detection; Plant leaves; Deep Learning; Convolutional Neural Network; Transfer learning; Agriculture.

        · abstract in Spanish     · text in English     · English ( pdf )