Field-captured dataset of banana plants across growth stages and early disease conditions
Description
The dataset comprises of 1,090 high-resolution images of banana plants that were collected under natural lighting conditions in the field with a digital camera (smartphone) from a plantation in Solapur, Maharashtra, India. The underlying research hypothesis for the collection is that a sufficiently diverse collection of raw field images will provide a basis for developing reliable computational vision and machine learning tools for early detection of plant diseases and growth stage analysis of banana crops. The banana plants shown in these photo images were collected at different growth stages and quality, including saplings, raw bananas with flower, fully grown healthy plants, and plants showing early recognizable signs of disease with visible leaf spots, discoloration, or wilting. The photo images are provided in raw format, left unprocessed and are stored in JPG format with variable pixel resolutions. By purposefully leaving the dataset unlabeled and unfiltered, the full pre-processing, annotating, and structuring of the data for the user's research is at the users discretion, so as to limit curator bias. There are several unique features of this dataset including its authenticity (being collected directly in the field), size (over 1,000 images), and richness (different growth stages, healthy and diseased).The dataset can be interpreted and reused in multiple ways, such as for training deep learning models for disease classification, evaluating growth stages in crop monitoring systems, or serving as benchmark data for precision agriculture applications. Ultimately, this dataset is intended to support research that contributes to sustainable farming practices through timely disease detection and improved crop management.