On this page
Research Article | Open Access
Volume 14 2022 | None
Cow Disease (LSD) Classification System for predicting different Severity levels.
V.Sivamurugan, K.R. Uthayan, and V. Thanikachalam
Pages: 2860-2873
Abstract
Lumpy Skin Disease is identified as a major threat to cattle production. It has a substantial impact on livelihoods and food security globally across the different countries in the world. Large skin nodules covering all parts of the body, fever, nasal discharge, spread lymph nodes and lacrimation are identified as the symptoms for this Lumpy skin disease. The virus responsible for this disease may spread due to direct contact to the skin lesions, saliva, nasal discharge, milk, or semen of cattle/animals infected by this disease. Unfortunately, there are no specific antiviral drugs or medicines available so far for the control and treatment of this Lumpy Skin Disease. The only mechanism available for control and treatment is the supportive care of cows. Thus, early detection helps to provide treatment before reaching the abnormal conditions. However, manual diagnosis and detection takes a significant amount of time and requires a trained professionally qualified and experienced person. Therefore, Artificial Intelligence based technology is needed to prevent and stop the animal disease epidemics. In this research work, we have proposed a classification system based on deep learning for predicting the different severity levels or grades of Lumpy Skin Disease. The proposed system uses a dataset of images of LSD-infected cattle, which have been labelled based on severity as mild, severe, and normal level by Veterinary doctors. We have utilized different pre-trained convolutional neural network (CNN) models customized by adding new layers for training by using the features extracted from the images. The trained model is then used to predict the severity levels of new Lumpy Skin Disease infected cattle images. We have evaluated the performance of our proposed system on a test dataset and achieved a high accuracy rate of 0.9182%. We have also compared our results across different models such as VGG19, Inception V3, Xception, ResNet50, DenseNet121 and MobileNetV2 and have demonstrated the superiority of our approach. Our proposed system has the potential to be used as a diagnosing tool for early detection and classification of LSD in cattle, enabling dairy farmers to take appropriate measures to prevent the spread of the disease and alleviate its impact.
Keywords
Augmentation, Classifier model, Lumpy Skin Disease, Pre-trained CNN models, Mobile Net V2
PDF
32
Views
4
Downloads