Rehan Sarwar1*, Muhammad Aslam1, Khaldoon S Khurshid1, Tauqir Ahmed1, Ana Maria Martinez-Enriquez2 and Talha Waheed1
1Department of Computer Science, University of Engineering and Technology, Lahore, Pakistan 4Arid Agriculture University, Rawalpindi, Pakistan 5University of Haripur, Pakistan
2Department of Computer Science, Center of Investigations and Advanced Studies (CINVESTAV-IPN), Mexico City, Mexico
1Department of Computer Science, University of Engineering and Technology, Lahore, Pakistan
4Arid Agriculture University, Rawalpindi, Pakistan
5University of Haripur, Pakistan
*Corresponding Author: Rehan Sarwar, Department of Computer Science, University of Engineering and Technology, Lahore, Pakistan.
Received: July 28, 2021; Published: September 24, 2021
Citation: Rehan Sarwar., et al. “Detection and Classification of Cotton Leaf Diseases Using Faster R-CNN on Field Condition Images". Acta Scientific Agriculture 5.10 (2021): 29-37.
Disease’s classification of the cotton leaf can increase the cotton yield. Deep learning is emerging as a powerful method in many fields, and much research has been done in Agriculture in real-time detection of cotton leaf diseases. Convolution Neural Networks have played a vital role in plant classification and identification of diseases, but still, work is needed to help farmers and pathologists correctly detect and classify diseases. Manual checking of disease crops takes a lot of time and cost, and it is hectic. Besides, the wrong diagnosis entails inaccurate conclusions, treatment, and significant expense. This paper proposes to train a deep learning Faster R-CNN model on cotton crop leaf dataset (CCLDataset) for detecting and classifying diseases on leaves, including both healthy and diseases. Plant Village dataset is the reference in finding the best feature extractor from VGG-16, InceptionV1, and V2. Additionally, it is a base model in Faster R-CNN. Transfer learning is performed on CCLDataset when trained on the model Faster R-CNN inceptionV2 coco by replacing the output layers of coco with CCLDataset to detect and classify leaf diseases. The experimental results show a mean average precision (mAP) of 87.1%.
Keywords: Cotton Leaf Diseases; Object Detection; Faster R-CNN; CCLDataset; Mean Average Precision
Copyright: © Rehan Sarwar., et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.