scaffora

segmentation loop run for the first time; sometimes it works sometimes it doesn’t?? going to try otsu thresholding next

using thresholding segmentation since its the simplest to use!! i want to get rid of the scale label on each of the images in order to only measure features from the coral branch itself using contouring

segmentation & masking
i’m learning how to do segmentation with SAM & cv2 in order to isolate the coral branching structures. the structures will be used to measure things like branching angle, thickness, ‘porosity,’ etc.

i realized that sam was not able to isolate from images where the coral had too much overlap, so i replaced the dataset with simpler, 1-piece-of-coral images. for more accuracy, i will be using point prompting

and so it begins.. i’ve uploaded 21 images for the initial dataset (coralsoftheworld), using different species for each architecture type. all are under the genus acropora, images are labeled by species name :) next i’ll be working on a segmentation function to isolate the coral from the background