Browsing by Author "Mbelwa, Hope"

Now showing 1 - 4 of 4

Deep Convolutional Neural Network for Chicken Diseases Detection
(International Journal of Advanced Computer Science and Applications, 2021) Mbelwa, Hope; Machuve, Dina; Mbelwa, Jimmy
For many years in the society, farmers rely on experts to diagnose and detect chicken diseases. As a result, farmers lose many domesticated birds due to late diagnoses or lack of reliable experts. With the available tools from artificial intelligence and machine learning based on computer vision and image analysis, the most common diseases affecting chicken can be identified easily from the images of chicken droppings. In this study, we propose a deep learning solution based on Convolution Neural Networks (CNN) to predict whether the faeces of chicken belong to either of the three classes. We also leverage the use of pre-trained models and develop a solution for the same problem. Based on the comparison, we show that the model developed from the XceptionNet outperforms other models for all metrics used. The experimental results show the apparent gain of transfer learning (validation accuracy of 94% using pretraining over its contender 93.67% developed CNN from fully training on the same dataset). In general, the developed fully trained CNN comes second when compared with the other model. The results show that pre-trained XceptionNet method has overall performance and highest prediction accuracy, and can be suitable for chicken disease detection application.
Deep learning-based mobile application for the enhancement of pneumonia medical imaging analysis: A case-study of West-Meru Hospital
(Elsevier, 2024-09) Kimeu, Japheth; Kisangiri, Michael; Mbelwa, Hope; Leo, Judith
Pneumonia remains a significant global health challenge, demanding innovative solutions. This study presents a novel approach to pneumonia diagnosis and medical imaging analysis, leveraging advanced technologies. The study used a Literature Review Methodology to study various scientific articles and involved healthcare staff, including Doctors, Nurses, Radiologists and the community, in sharing their requirements for the study. The findings led to the proposal for the integration of Deep Learning techniques, including Convolutional Neural Network (CNN), as well as tools like YOLOv8, Roboflow, and Ultralytics, to revolutionize pneumonia detection and classification. The EfficientDet-Lite2 model architecture was subsequently used to generate a TensorFlow Lite Model, deployable in both Android and iOS mobile applications. The study’s outcomes reveal a substantial improvement in the precision and recall metrics. These results signify a promising step forward in empowering healthcare professionals with timely and reliable results for optimal patient management.
Identification of Resistant Common Bean Genotypes to Foliar Diseases using Hyperspectral Data
(IEEE, 2023-11-15) Kaijage, Shubi; Leo, Judith; Mamo, Teshale; Mbelwa, Hope; Guerena, David; Irakiza, Josiane
Common bean is one of many legumes (family Fabaceae) widely cultivated for their edible seeds, seedpods, and leaves. Despite its benefits and life dependency especially in sub-Saharan Africa, production of improved common bean seeds is still a problem. Among the major problems that the bean breeders are facing is manual phenotyping which led to slowing the field processes.According to the literature, imaging technologies have been introduced to help in different processes for crop management and production. However, there is a lack of automated mechanisms for phenotyping processes. Therefore, this study has reviewed existing mechanisms among breeders in Tanzania in order to propose suitable mechanisms for effective phenotyping processes. In the process of data collection, the study interviewed 60 participants among which were common bean breeders and field technicians. Questionnaires and group discussion methods were deployed to understand the insight of the participants on the current. It was observed that manual phenotyping causes biases of data due to human judgments and leads to late decision-making. Additionally, the study reviewed existing literature and it was noted that the use of technologies such as spectrometry and remote sensing technologies were found to be suitable for early and continuous plant health monitoring. However, these technologies are of high cost and produce complex outputs for the breeders to understand. However, there is a way in which these techniques can be innovatively integrated with artificial intelligence (AI) techniques in order to produce required outputs that are easy to understand and cost-effective.Based on the findings, this study proposes the development of a Hyperspectral-based system that will be used for automating phenotyping, data collection, and analysis. The system will be developed using AI that uses spectral bands captured using a cheap handheld spectrometer for the early detection of foliar diseases in bean varietie...
Image - based poultry disease detection using deep convolutional neural network
(NM-AIST, 2021-07) Mbelwa, Hope
The poultry sector in the country is highly affected by diseases including Coccidiosis, Salmonella and Newcastle that have a significant impact on production. Lack of reliable information and proper methods of farming has led to the spread of the diseases as the majority of farmers practice traditional farming, hence lack a systematic way to detect and diagnose the disorders. Poultry farmers rely on experts to diagnose and detect the diseases; access to the experts is also a challenge due to the limited number of extension officers. With the available tools from artificial intelligence and machine learning, there is a potential to semi-automate the diagnostics process for the most common diseases in chickens. This study proposes a solution for predicting diseases in chickens using faecal images and deep Convolutional Neural Networks (CNN). Additionally, the work leverages the use of pre-trained models and develop the solution for the same problem. Based on the comparison, it is indicated that the model developed from the XceptionNet deep learning framework outperforms other models for all the metrics used. The experimental results indicate the accuracy of transfer learning at 94% using pre-training over the other models from fully training on the same dataset. The results show that the pre-trained XceptionNet framework (94%) has the best overall performance and highest prediction accuracy, and can be suitable for chicken disease detection application. The findings show that the proposed model is ideal for poultry diseases detection method.