A Unified Video Segmentation Benchmark: Annotation, Metrics and Analysis

By Fabio Galasso, Naveen Shankar Nagaraja, Tatiana Jimenez Cardenas, Thomas Brox, Bernt Schiele. 2013

175 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Unsupervised Learning of Edges (2015, methodology): “We attribute this to the small size of VSB.”
Playing for Benchmarks (2017, background): “The importance of video and motion cues is widely recognized [1, 18, 30, 49].”
YouTube-VOS: A Large-Scale Video Object Segmentation Benchmark (2018, methodology): “In comparison to previous datasets [16,21,22,23,24,25], DAVIS has both higher-quality of video resolutions and annotations.”
Click Carving: Segmenting Objects in Video with Point Clicks (2016, unknown): “Video segmentation propagation on VSB100: Next, we test on VSB100.”
Spectral Graph Reduction for Efficient Image and Streaming Video Segmentation (2014, methodology): “The algorithm of [13] (hierarchical image segmentation [2] propagated with optical flow) is only partially outperformed by the proposed methods.”
Optical Flow with Semantic Segmentation and Localized Layers (2016, background): “There is significant and increasing interest in the field [12, 14, 32, 41, 42, 57] but the definition of the problem varies between identifying coherent motions or coherent objects regions.”
Semantic Video Segmentation by Gated Recurrent Flow Propagation (2016, background): “In contrast, fully trainable approaches to semantic video segmentation would face the difﬁculty of obtaining detailed annotations for the individual video frames, although datasets are emerging for the (unsupervised) video segmentation problem[11].”
YouTube-VOS: Sequence-to-Sequence Video Object Segmentation (2018, methodology): “Even if we combine all the videos from available datasets [21,13,26,4,30,15], its scale is still far smaller than other video analysis datasets such as YouTube-8M [1] and ActivityNet [17].”
Learning Must-Link Constraints for Video Segmentation Based on Spectral Clustering (2014, methodology): “In the second set of experiments we consider the novel benchmark VSB100 [17], which includes 100 HD quality videos [41] arranged into train and test sets (40+60) (we split test – 25 – and validation set – 15).”
Fast action proposals for human action detection and search (2015, background): “However, these action proposal approaches highly rely on relatively accurate video segmentation [37, 38], which itself is a challenging problem.”
Deep Feature Flow for Video Recognition (2016, methodology): “The proposed method can easily be trained on datasets where only sparse frames are annotated, which is usually the case due to the high labeling costs in video recognition tasks [29, 11, 6].”
The 2019 DAVIS Challenge on VOS: Unsupervised Multi-Object Segmentation (2019, background): “[40], [41], [42] often interpreting the task in different ways.”
Track and Segment: An Iterative Unsupervised Approach for Video Object Proposals (2016, methodology): “The third type of methods, which is closest to our approach, compute segments on each frame and link them together through bottom-up or learned appearance matching and optical flow [6, 32, 3, 40, 16, 38, 53, 15].”
Robust video segment proposals with painless occlusion handling (2015, methodology): “Experiments We use the challenging VSB-100 dataset from [16] which contains 60 testing video sequences with pixel-level ground truth annotations every 20 frames.”