52 datasets found

The 2018 davis challenge on video object segmentation

RVOS-D provides more complex language descriptions from a broader object categories within relatively longer video sequences.

The dataset used for video object segmentation task, where a first frame labelled with the masks of several object instances, one aims to find the corresponding masks of objects...

The DAVIS-585 dataset is used for evaluation.

The dataset is used for video salient object detection and segmentation. It contains video saliency detection datasets, including DAVSOD, DAVIS, FBMS, SegV2, and ViSal.

The TSD-max dataset is a real-world traffic scene dataset used for video object segmentation.

This paper proposes the Davis 2017 validation set for video object segmentation.

A large-scale video object segmentation benchmark.

A benchmark for unsupervised multi-object segmentation.

A large-scale synthetic VOS dataset with pixel-accurate segmentations.

The MiniVSPW dataset is a benchmark for few-shot video object segmentation.

Few-Shot Video Object Segmentation (FSVOS) aims to segment objects in a query video with the same category defined by a few annotated support images.

The YouTube-VOS dataset is a sequence-to-sequence video object segmentation dataset, with 1000 videos and 1000 frames per video.

The paper proposes a framework for interactive video object segmentation (VOS) in the wild, where users can choose some frames for annotations iteratively.

The Point-VOS YouTube dataset consists of 4.4K videos with 4.4K objects annotated with points.

The Point-VOS DAVIS dataset consists of 32K videos with 600 objects annotated with points.

The Point-VOS Kinetics dataset consists of 23.9K videos with 120K objects annotated with points.

The Point-VOS Oops dataset consists of 8.4K videos with 13.1K objects annotated with points.

Point-VOS: Pointing Up Video Object Segmentation. Current state-of-the-art Video Object Segmentation (VOS) methods rely on dense per-object mask annotations both during training...

The DAVIS dataset contains 60 training sequences and 30 validation sequences, with high-quality densely-annotated segmentation mask annotation for each frame.