New procedure for protozoan white sport disease detection in mariculture fish based on artificial intelligence
Protozoan white spot disease, caused by Cryptocaryon irritans, poses a significant threat to marine fishes, resulting in a 100% mortality rate within a short period. Its impact extends beyond fish health, severely affecting commercial mariculture and causing substantial economic losses. The current...
Saved in:
| 主要作者: | |
|---|---|
| 格式: | Thesis |
| 語言: | English English |
| 出版: |
2024
|
| 主題: | |
| 在線閱讀: | https://etd.uum.edu.my/11301/1/depositpermission.pdf https://etd.uum.edu.my/11301/2/s828023_01.pdf https://etd.uum.edu.my/11301/ |
| 標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
| 總結: | Protozoan white spot disease, caused by Cryptocaryon irritans, poses a significant threat to marine fishes, resulting in a 100% mortality rate within a short period. Its impact extends beyond fish health, severely affecting commercial mariculture and causing substantial economic losses. The current standard procedure for early fish disease detection, reliant on time-consuming external gross observation, underscores the need for faster and more reliable automated approaches. In response to this issue, this study explores Artificial Intelligence (AI) as a transformative tool in fish disease detection. This study involved fine-tuning several convolutional neural networks (CNN) architectures including AlexNet, ResNet50, GoogleNet, ResNet101, and Vgg16Net to discern the most effective architecture for the proposed new procedure. Leveraging the success of CNN, particularly using the CNN’s ResNet50 architecture, this study introduced a new procedure via integration of contrast-adaptive colour correction (CACC) with CNN for protozoan white spot disease detection using underwater images. Experimental finding reveals outstanding performance of the procedure, with the ResNet50 CNN architecture integrated with CACC achieving a testing accuracy of 99.52% on the acquired dataset, indicating that this new procedure holds immense promise in early fish disease prevention and intervention by providing an efficient and precise approach to protozoan white spot disease detection. While this experimental stage study demonstrates encouraging results, future endeavours should focus on extensive refinement and collaboration to elevate the procedure’s readiness for on-site implementation. In summary, the integration of AI-driven techniques not only improves detection accuracy but also streamlines processes, potentially safeguarding marine fish populations and the aquaculture industry. |
|---|