See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/382562684 House plant flower disease detection using deep learning Conference Paper · July 2024 CITATIONS 0 READS 460 4 authors , including: Nisal Thiwanka Uva Wellassa University 7 PUBLICATIONS 8 CITATIONS SEE PROFILE All content following this page was uploaded by Nisal Thiwanka on 26 July 2024. The user has requested enhancement of the downloaded file. 8 th International Research Conference of Uva Wellassa University, IRCUWU2024 “Sustainability Nexus: Multidisciplinary Connections for a Resilient Future” July 24 - 25, 2024 @ Uva Wellassa University, Badulla, Sri Lanka Paper ID: IRCUWU2024-580 Oral House plant flower disease detection using deep learning T.R. Bandara Q , W.M.M. Weerakkodi, C. Subadharshini, M.N.T. Nandasena Department of Computer Science and Informatics, Uva Wellassa University, Badulla, Sri Lanka Q iit19014@std.uwu.ac.lk, +94761702791 Incorporating flowers into our life can be a wonderful way to minimize stress and create a more peaceful and balanced environment, even with a busy schedule. These plants, cultivated for their ornamental value, contribute to interior aesthetics while offering a touch of nature within living spaces. Apart from that, if they are able to grow flowers without diseases, these growers can also export flowers to foreign markets. Developing strategies for early detection of plant diseases offers the advantage of increasing flower yields and reducing reliance on pesticides regardless of the specific disease causing the problem. The study aimed to help farmers identify and diagnose diseases affecting Roses, Anthurium and Gerberas. Developing strategies for early detection of flower diseases can significantly increase both yield and quality of domestic flower production. This research employed image-based prediction techniques to develop three separate models for detecting diseases in Roses, Anthuriums, and Gerbera using image processing, deep learning, and machine learning. It is capable of accurately detecting diseases in these ornamental flowers. The research endeavored to integrate these models into a user-friendly mobile application, enabling flower growers to quickly and accurately diagnose plant diseases, thereby enhancing disease management practices. The dataset we created contained two diseases of rose plants, two diseases of anthurium plants and two diseases of gerbera plants. 690 sample images were used for Roses. For roses keras framework and transfer learning were used and modify a model using MobileNet as a base, typically load the MobileNet architecture with pretrained weights and modify it according to specific requirements. 100% accuracy was achieved for rose. 200 sample images were used for Gerbera and keras was used for modify the model. 97.5% accuracy was achieved for Gerbera. 120 sample images were used for Anthurium and used for TensorFlow object detection pre trained model. The research holds promise for improving the aesthetic and economic value of indoor plants by providing timely and accurate disease detection, ultimately contributing to the fields of ornamental horticulture and plant pathology. Keywords: Flowers; disease detection; image processing; ornamental horticulture Underlined is the presenting author. 85 View publication stats