Datasets
From Computer Vision at Waterloo
We host the following datasets freely available for download.
- Synthetic Vessel Data 15 volumes of synthetic 3D vessel tree data that include CT-like volume (w/ std10 and std15 additive Gaussian noise respectively) and ground-truth vessel centerline tree.
- Modified version of code and evaluation metrics: here
- Relevant papers: Z. Zhang 2019G. Hamarneh 2010.
- Convexity Database A collection of 51 natural images with convex or almost convex objects. The database includes original images, user scribbles and segmentation results using our Convexity Trust Region method.
- Max-flow problem instances in vision corresponding to segmentation, stereo, multiview reconstruction, and other problems in computer vision, graphics, and biomedical imaging. Instances are in DIMACS format and are useful for benchmarking max-flow/min-cut algorithms.