BB-HSI: Proximal Hyperspectral Imaging Dataset for Masena Blueberries Epicuticular Wax Loss Identification and Post-Harvest Quality Evaluation.

We present a close-range hyperspectral image dataset of 39 unique Masena blueberry fruits harvested by different methods for epicuticular wax (EW) preservation and post-harvest quality studies. The dataset is intended to support close-range hyperspectral imaging research in agricultural and food analysis, particularly in the evaluation of harvesting techniques and the effect on surface properties like wax layers. Hyperspectral data were captured using the Specim FX17e camera, with a spectral range of 900–1700 nm in 224 spectral bands, under halogen illumination. The hyperspectral cubes include spatial (1046x640) and spectral (224) data of blueberry samples placed in 3D-printed PLA trays. Berries were harvested mid-season on November 24, 2023, from an orchard at Pukehina, New Zealand, where tunnel coverings protected the fruit from environmental agents degrading EW. Three harvest conditions were studied: (i) Control: berries picked by hand carefully with polyvinyl chloride (PVC) gloves; (ii) Hand-harvested (HH): berries hand-harvested conventionally; and (iii) Assisted harvested (AH): berries harvested using a custom-built handheld shaker designed at the University of Waikato. The samples were maintained under controlled environments and imaged within 9 hours of harvest. The images were annotated with the authors' in-house developed HAPPy (Hyperspectral Application Platform in Python) tool (ENVI Software) and converted into MATLAB (.mat) files for analysis using machine/deep learning models. A total of 49 unique hyperspectral images were acquired from 39 blueberry fruits to capture multiple views or surface conditions. We present 5 sets of spectral hypercubes collected by the HSI camera: ‘Assisted Harvested Blueberries (AHB)’ (10 images), ‘Hand Harvested Blueberries (HHB)’ (10 images). ‘No EW’ (9 images), ‘Perfect EW’ (10 images), and ‘No EW vs. Perfect EW’ (10 images: 5 from No EW and 5 from Perfect EW). This dataset, collected and maintained by the University of Waikato (WaI2M: Waikato Instrumentation and Measurement Research Group & Hyperspectral Imaging Group), can be used for tasks such as EW degradation detection, classification of harvesting methods, and spectral analysis of surface properties in fruits using different machine learning and deep learning models.