Soul: An OCTA dataset based on a human-machine collaborative annotation framework

The data includes raw images, corresponding labels, and clinical text information sheets. Soul.zip includes six different subsets, each corresponding to a different number of surgeries and follow-ups, corresponding to the names s1, s2, etc. Each subset contains the original image and the labels generated by the human-robot collaboration framework. An Excel worksheet contains two sub-tables, patient-level details and individual picture-level details.