Gloss is an important characteristic in the quality evaluation in chocolate production. However, standard glossing measuring devices face several challenges when measuring food products, in particular those with curved surfaces and small, such as chocolate sprinkles. Therefore, gloss evaluation of chocolate sprinkles is typically done by human visual assessment. In this respect, hyperspectral imaging (HSI), combining spectroscopy and imaging, has gained attention as a non-destructive and non-contact real-time detection tool for food quality analysis and control. This technique adds an extra dimension to traditional machine vision techniques by providing images at a larger number of more narrow wavebands. This can potentially increase the discrimination power. The main task of this dataset is to classify between 5 classes of chocolate production stages. The labels are indicated on the file names. The labels are: EXTRUDER, GLUCOSE, STAGE1, STAGE2, STAGE3. The .zip file contains two folders named: - envi_sprinkles_dataset_11_04_22 (Train Set) - envi_sprinkles_test_dataset_11_04_22 (Test Set) In envi_sprinkles_dataset_11_04_22 the file name convention is BATCH_CLASSNAME_INDEX.hdr In envi_sprinkles_test_dataset_11_04_22 the file name convention is CLASSNAME_INDEX.hdr Note that in the train set the batch information is present in the file name, if two samples belongs to the same batch means that they were produced at the same time. The samples format is ENVI. Consist of a header file and ENVI binary data file with file extensions .hdr and .raw, respectively. The function writes the wavelength and metadata information to the ENVI header file and the data cube containing the hyperspectral images to the ENVI binary data file. The image shape is [272x512x16] , [HEIGHT, WIDTH, BANDS], the pixel format is uint16.