OVIS Occluded Video Instance Segmentation Dataset

OVIS (a large-scale dataset for occluded video instance segmentation)Occluded Video Instance Segmentation ), which can be used to detect, segment, and track instances in occluded scenes.

OVIS contains 296,000 high-quality mask annotations of 5,223 target objects, covering 25 common categories in life, including: people, birds, cats, dogs, horses, sheep, cows, elephants, bears, zebras, giraffes, poultry, giant pandas, lizards, parrots, monkeys, rabbits, tigers, fish, turtles, bicycles, motorcycles, airplanes, ships and vehicles.

The average video duration and average instance duration are 12.77s and 10.55s respectively.

The dataset was proposed by a team from Huazhong University of Science and Technology, Alibaba, Cornell University, Johns Hopkins University, and Oxford University.

Related paper: Occluded Video Instance Segmentation

Paper address: https://arxiv.org/abs/2102.01558