ShrimpDiseaseImageBD: An Image Dataset for Computer Vision-Based Detection of Shrimp Diseases in Bangladesh
Description
The ‘ShrimpDiseaseImageBD’ is a dataset of images designed to help in the identification of shrimp diseases commonly found in aquaculture within Bangladesh using computer vision techniques. This dataset has the potential to play an important role in the aquaculture field, where detection of the shrimp disease with proper timing and precision are very important to monitor shrimp health and maintain economic security of farmers. The ShrimpDiseaseImageBD’ dataset contains images of different shrimp diseases collected in collaboration with local shrimp farms. The dataset will assist in developing and evaluating automated disease detection models and will prove to be of value to anyone working in the field of computer vision and aquaculture. Enriched with images of shrimp diseases, such a dataset holds significant potential to stimulate much research and practical application in automated health monitoring in aquaculture. The dataset is released to the public domain to encourage further research and development in this domain. The dataset is systematically organized into two main directories within the "Root" folder: "Raw Images" and "Annotated Diseased Shrimp Images." The "Raw Images" directory contains 1149 images, divided into four subfolders: "Healthy" (403 images), "BG" (198 images), "WSSV" (328 images), and "BG_WSSV" (220 images). Each folder categorizes the images according to the health status or disease type, providing a structured arrangement for easy access. In the "Annotated Diseased Shrimp Images" directory, the images are organized into three subfolders named "BG," "WSSV," and "BG_WSSV", each containing two additional subfolders named "Images" and "Labels." The "Images" subfolder includes the actual shrimp images, while the "Labels" subfolder holds annotation data for the corresponding images. This structured organization facilitates efficient navigation and supports effective use in machine learning model training by providing both raw and annotated data in a clear and accessible format.