Supplemental material

The folders contain the raw monochromatic images obtained with the hyperspectral system and reported in the paper. Each set of images is ordered from 420nm to 730nm in 5nm steps. <br> The folder "Optical Signatures (Training Set)" includes all the images used to construct hypercubes and from which ROIs were selected and labelled to create the spectral library used for training. The .txt file "ROIs" inside each one of these folders contains the ROI selected from the original images (cropping step of pre-processing); the .txt file "ROI_MATLAB" contains the rectangular ROIs from which pixels were collected to create the training data set; the .txt file "ROI_labels_MATLAB" contains the respective labels. <br> Similarly, the "Model Target (Test Set)" folder contains the images of the spatially disjoint model target used for test/validation of the model. It also contains the ROIs and respective labels in .txt files used to create the two test datasets (for the fine and coarse classification). This is the hypercube presented in Figure 7. <br> The folder "Calibration" contains the Bright Field and Dark Field images used for pixelwise calibration according to Equation (2), allowing to convert the intensity data to global reflectance. <br> Finally, the raw data for the real targets with biofouling is also included (the corresponding images in the manuscript are disclosed in the filename). <br> <br> <br>