Please use this identifier to cite or link to this item: https://gnanaganga.inflibnet.ac.in:8443/jspui/handle/123456789/15411
Full metadata record
DC FieldValueLanguage
dc.contributor.authorReddy, Boreddy Sai Yaswanth-
dc.contributor.authorDileep, N-
dc.contributor.authorShrivastava, Virendra Kumar-
dc.date.accessioned2024-04-20T10:53:13Z-
dc.date.available2024-04-20T10:53:13Z-
dc.date.issued2023-
dc.identifier.urihttp://gnanaganga.inflibnet.ac.in:8080/jspui/handle/123456789/15411-
dc.description.abstractPlant diseases pose a significant threat to global food security, as they can causesubstantial losses in crop yields and lead to economic damages for farmers and the agricultural industry. Traditional methods of plant disease detection, which rely on human experts' visual observations, can be time-consuming, expensive, and prone to errors. As such, there is a need for more efficient and accurate methods of detecting plant diseases. Deep learning models offer several advantages over traditional methods of plant disease detection. These models can process large amounts of data quickly and accurately, identifying patterns that may not be visible to human experts. To train and test the proposed models, the study uses images from the publicly available plant village dataset, which consists of over 50,000 images belonging to 14 different plant species. Additionally, the dataset will be augmented through image augmentation techniques to ensure a diverse and robust representation ofplant leaf diseases. The models built were trained with 8 different plant species Apple, Tomato, Grape, Potato and Strawberry, corn, cherry and pepper. A total of23074 images were collected from plant village dataset belonging to 8 plant species. A total of 16 diseases were classified along with healthy leaves belongingto those eight species. To address this need, this study proposes multiple deep learning models based onXception and Densenet. The first model combines the depth-wise separable convolutions of Xception with the feature reuse capabilities of Densenet, while thesecond model concatenates both models to obtain a new model that trains on bothsimultaneously. The two models are then concatenated using a global average pooling (2D) layer. The proposed models were trained with and without data augmentation, and the study compared the models' performance based on accuracy, robustness, generalization capability, training time, memory consumption, and various other parameters. To make the system more accessible to end-users, the study also developed a user-friendly interface that allows non-experts to use the system to detect diseases in their plants, without the need for specialized training or expertise in deep learning.en_US
dc.language.isoenen_US
dc.publisherAlliance College of Engineering and Design, Alliance Universityen_US
dc.subjectPlant Leaf Diseaseen_US
dc.subjectDeep Learningen_US
dc.subjectPlant Disease Detectionen_US
dc.subjectPooling (2D) Layeren_US
dc.titlePlant Leaf Disease Detection Using Deep Learningen_US
dc.typeOtheren_US
Appears in Collections:Dissertations - Alliance College of Engineering & Design

Files in This Item:
File SizeFormat 
CSE_G19_2023.pdf
  Restricted Access
1.46 MBAdobe PDFView/Open Request a copy


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.