Basil Disease and Variety Recognition Dataset

The database for the current project was created based on primary data collection related to basil leaves (Ocimum basilicum). The process of primary data collection was conducted in cooperation with the Biotechnology Department and the herbal garden located within the university campus. In particular, the main purpose of the process of data collection was to provide a reliable database of healthy and unhealthy basil leaves that were cultivated under certain conditions.
During the process of primary data collection, regular trips to the herbal garden were made, and the plants were inspected for signs of diseases. The identification of healthy and unhealthy leaves was facilitated by biotechnology specialists. To represent a more varied image database reflecting different agricultural environments, high-quality images were captured with the help of the camera of a smartphone. Different light conditions, viewing angles, and backgrounds were used for photo shooting. Disease symptoms observed during the process of data collection were also recorded for further multimodal analysis.
After the collection, the acquired images were further screened and preprocessed. Blurry, redundant, and images with poor quality were discarded during preprocessing. After careful screening, experts grouped the images into their respective disease classes for easy access and classification. Other metadata about the images such as disease class, location of collection, etc., were also stored.
Deposition was done at the Mendeley Data archive site for easy access, sharing, transparency, and collaborations in future research. Mendeley Data is a well-known open-access platform for storing and sharing research datasets. Using the Mendeley platform, our dataset was archived in a secure repository where it can be permanently preserved and cited with the assigned DOI.
Making the dataset publicly available can help other researchers validate experimental results, benchmark machine learning models, or train more effective plant disease detection systems. Additionally, sharing data openly can accelerate science by promoting open science practices, fostering collaboration, and avoiding unnecessary replication of data collection. The dataset might also be used as educational material for students and scientists interested in agriculture, biotechnology, computer vision, or artificial intelligence.
As we upload our basil leaf dataset to Mendeley Data, we believe that this research presents a reliable disease detection framework as well as publicly available dataset that could be leveraged for future smart agriculture and plant monitoring research. Collaboration with the herbal garden allowed us to collect authentic dataset which was further verified by the Biotechnology Department for quality.