The 2019 DAVIS Challenge on VOS: Unsupervised Multi-Object Segmentation

By Sergi Caelles, Jordi Pont-Tuset, Federico Perazzi, Alberto Montes, Kevis-Kokitsi Maninis, Luc Van Gool. 2019

179 citations recorded by the source corpus.

We present the 2019 DAVIS Challenge on Video Object Segmentation, the third edition of the DAVIS Challenge series, a public competition designed for the task of Video Object Segmentation (VOS). In addition to the original semi-supervised track and the interactive track introduced in the previous edition, a new unsupervised multi-object track will be featured this year. In the newly introduced track, participants are asked to provide non-overlapping object proposals on each image, along with an identifier linking them between frames (i.e. video object proposals), without any test-time human supervision (no scribbles or masks provided on the test video). In order to do so, we have re-annotated the train and val sets of DAVIS 2017 in a concise way that facilitates the unsupervised track, and created new test-dev and test-challenge sets for the competition. Definitions, rules, and evaluation metrics for the unsupervised track are described in detail in this paper.

Subfields: Computer Vision and Pattern Recognition, Evaluation & Benchmarks

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

In Evaluation & Benchmarks

Citations in the corpus

High-Resolution Image Inpainting with Iterative Confidence Feedback and Guided Upsampling (2020, methodology): “al image editing scenario, in this work, we synthesize training samples with realistic holes. We collect 82,020 object masks from densely annotated segmentation datasets, including video segmentation [11], semantic segmentation [18], salient object segmentation [27,35,19], and human parsing [28]. Th”
SAM 2: Segment Anything in Images and Videos (2024, methodology): “After pre-training, we train SAM 2 on our introduced datasets SA-V + Internal (section §5.2), a 10% subset of SA-1B, and a mixture of open-source video datasets including DAVIS (Pont-Tuset et al., 2017; Caelles et al., 2019), MOSE (Ding et al., 2023), and YouTubeVOS (Xu et al., 2018b).”
Conditional Object-Centric Learning from Video (2021, methodology): “We used the open source implementation provided by the authors to train their model on our datasets, MOVi and MOVi++; and evaluated tracking performance using the Davis 2017 benchmark utilities (Caelles et al., 2019; Pont-Tuset et al., 2017) on the validation sets of MOVi and MOVi++ respectively.”
Learning Video Object Segmentation From Unlabeled Videos (2020, background): “The effectiveness of instance-level Z-VOS can be observed in tram of DAVIS 17 [4].”
Opening up Open World Tracking (2021, methodology): “Closer to our work is unsupervised video object segmentation (UVOS) [13] and motion segmentation [9, 30, 71], where multiple objects that are present throughout the video and exhibit dominant motion need to be tracked and segmented.”
Video Instance Segmentation 2019: A Winning Approach for Combined Detection, Segmentation, Classification and Tracking. (2019, background): “Another task related to VIS is Multi-Object Unsupervised Video Object Segmentation (UVOS) [1].”
STEm-Seg: Spatio-temporal Embeddings for Instance Segmentation in Videos (2020, methodology): “More relevant to our work is the task of unsupervised VOS [10,68].”
A Survey on Deep Learning Technique for Video Segmentation (2021, methodology): “Moreover, as the original annotations of DAVIS17 are biased towards the SVOS scenario, DAVIS19 challenge [177] reannotates val and test-dev sets of DAVIS17 to support AVOS.”