ORVS (Online Retinal image for Vessel Segmentation (ORVS))

Introduced by Sarhan et al. in Transfer Learning Through Weighted Loss Function and Group Normalization for Vessel Segmentation from Retinal Images

The ORVS dataset has been newly established as a collaboration between the computer science and visual-science departments at the University of Calgary.

This dataset contains 49 images (42 training and seven testing images) collected from a clinic in Calgary-Canada. All images were acquired with a Zeiss Visucam 200 with 30 degrees field of view (FOV). The image size is 1444×1444 with 24 bits per pixel. Images and are stored in JPEG format with low compression, which is common in ophthalmology practice. All images were manually traced by an expert who a has been working in the field of retinal-image analysis and went through training. The expert was asked to label all pixels belonging to retinal vessels. The Windows Paint 3D tool was used to manually label the images.

Source: Transfer Learning Through Weighted Loss Function and Group Normalization for Vessel Segmentation from Retinal Images

Papers


Paper Code Results Date Stars

Tasks


Similar Datasets


License


  • Unknown

Modalities


Languages