JavaScript is disabled for your browser. Some features of this site may not work without it.
| dc.contributor.author | EMIYAMREW, AZMERAW | |
| dc.date.accessioned | 2024-05-20T06:27:03Z | |
| dc.date.available | 2024-05-20T06:27:03Z | |
| dc.date.issued | 2023-07-17 | |
| dc.identifier.uri | http://ir.bdu.edu.et/handle/123456789/15792 | |
| dc.description.abstract | The therapeutic nature of medicinal plants and their ability to heal many diseases raises the need for their automatic identification. Consequently, automatic medicinal plant identification system was proposed using different neural network techniques in both machine learning and deep learning methods by numerous researchers. In medicinal plant identification systems, computational resource requirements should be considered to deploy it in mobile applications and to be used by farmers, botanists, junior traditional healers, and an individual easily. However, mobile devices have limited computational resources. So to afford this requirement, special attention should be paid to reducing the number of parameters and model complexity. This work explores deep learning and proposes a deep learning-based approach for the identification of medicinal plants using efficient and light weighted MobileNetV2 feature. The medicinal leaf datasets consists of 27 classes of 5300 Ethiopian indigenous medicinal plant leaves. The transfer learning approach along with the pre-trained neural networks such as VGG16, VGG19, InceptionV3, Xception, and MobileNetV2 architectures were employed to extract features from the input leaf images. The MobileNetV2 architecture outperformed the state-of-the-art pre-trained models by achieving 99.15% average validation accuracy of Support Vector Machine (SVM) and softMax classifier). The Artificial Neural Network (ANN) was used to learn the features from the proposed feature extractor MobileNetV2 model. To improve performance and reduce the complexity and over-fitting of the proposed model, the ANN classifier was tuned using the Bayesian optimization technique. Finally, the proposed model was validated with our custom datasets and achieved 99.62% classification accuracy. Furthermore, MobileNetV2/ANN was tested on two benchmark datasets and achieved 99.9% in flavia-plant and 99.7% on medicinal plant leaf datasets The finding of this work showcases transfer learning-based MobileNetV2 feature with Artificial Neural Network(ANN) classifier, can effectively identify medicinal plants from their leaf images with high accuracy of 99.62% and offers a promising solution to the fading traditional medicinal knowledge. Key words: Medicinal plant, transfer learning, deep learning, Artificial Neural Network. | en_US |
| dc.language.iso | en_US | en_US |
| dc.subject | Electrical and Computer Engineering | en_US |
| dc.title | EthioMedLeafNet: A Deep Learning Approach for Medicinal Plant Identification Using MobileNetV2 Features | en_US |
| dc.type | Thesis | en_US |
