Benchmark Almanac

DAVIS 2017

STM 72.2% tops its leaderboard
Not in model cards

Models track and segment individual objects throughout a video. The evaluation measures region overlap and boundary accuracy.

Released
April 2017
Kind
Video understanding
Level
Everyday
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English
In and out
Video, Images → Video
Interaction
Single turn
Run by
Anyone
20
Entries on its own leaderboard, 0 models we track

STM tops the leaderboard with 72.2%

Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

  1. 1 STM Semi-supervised, test-dev 72.2%
  2. 2 PReMVOS Semi-supervised, test-dev 71.6%
  3. 3 MHP-VOS Semi-supervised, test-dev 69.5%
  4. 4 CINM Semi-supervised, test-dev 67.5%
  5. 5 Lucid Semi-supervised, test-dev 66.7%
  6. 6 FEELVOS Semi-supervised, test-dev 57.8%
  7. 7 OSVOS-S Semi-supervised, test-dev 57.5%
  8. 8 RANet Semi-supervised, test-dev 55.4%
  9. 9 Siam R-CNN Semi-supervised, test-dev 53.3%
  10. 10 OnAVOS Semi-supervised, test-dev 52.8%
  11. 11 RGMP Semi-supervised, test-dev 52.8%
  12. 12 AGAME Semi-supervised, test-dev 52.3%
  13. 13 OSVOS Semi-supervised, test-dev 50.9%
  14. 14 RVOS Semi-supervised, test-dev 50.3%
  15. 15 AGS Unsupervised, test-dev 45.6%
  16. 16 FAVOS Semi-supervised, test-dev 43.6%
  17. 17 SIAMMASK Semi-supervised, test-dev 43.2%
  18. 18 MuG-W Unsupervised, test-dev 41.7%
  19. 19 OSMN Semi-supervised, test-dev 41.3%
  20. 20 PDB Unsupervised, test-dev 40.4%

Source: Official DAVIS 2017 leaderboard

#EntryScoreSettingDateSource
1 STM 72.2% Semi-supervised, test-dev Official leaderboard
2 PReMVOS 71.6% Semi-supervised, test-dev Official leaderboard
3 MHP-VOS 69.5% Semi-supervised, test-dev Official leaderboard
4 CINM 67.5% Semi-supervised, test-dev Official leaderboard
5 Lucid 66.7% Semi-supervised, test-dev Official leaderboard
6 FEELVOS 57.8% Semi-supervised, test-dev Official leaderboard
7 OSVOS-S 57.5% Semi-supervised, test-dev Official leaderboard
8 RANet 55.4% Semi-supervised, test-dev Official leaderboard
9 Siam R-CNN 53.3% Semi-supervised, test-dev Official leaderboard
10 OnAVOS 52.8% Semi-supervised, test-dev Official leaderboard
11 RGMP 52.8% Semi-supervised, test-dev Official leaderboard
12 AGAME 52.3% Semi-supervised, test-dev Official leaderboard
13 OSVOS 50.9% Semi-supervised, test-dev Official leaderboard
14 RVOS 50.3% Semi-supervised, test-dev Official leaderboard
15 AGS 45.6% Unsupervised, test-dev Official leaderboard
16 FAVOS 43.6% Semi-supervised, test-dev Official leaderboard
17 SIAMMASK 43.2% Semi-supervised, test-dev Official leaderboard
18 MuG-W 41.7% Unsupervised, test-dev Official leaderboard
19 OSMN 41.3% Semi-supervised, test-dev Official leaderboard
20 PDB 40.4% Unsupervised, test-dev Official leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Audio and video

All 695 in Audio and video
  • Video-MMMU In use 300 expert-level lecture videos with questions testing perception, comprehension and adaptation of knowledge. Accuracy. Qwen 3.8 Max 88.7%
  • LVBench In use Hours-long video understanding with multiple-choice questions on reasoning, grounding and summarization. Gemini 4 Argon 91.7%
  • MMVU In use 3,000 expert-annotated questions on videos across 27 disciplines. Multiple-choice and open-ended accuracy. Qwen 3.8 Max 82.4%
  • MLVU In use Multi-task long video understanding (minutes to hours) with multiple-choice and generation tasks. Accuracy. Qwen 3.8 Max 90.8%
  • FLEURS In use Models transcribe read speech in 102 languages built from FLoRes-101 sentences. Model cards usually report word error rate, so lower scores are better. Phi-4-multimodal 4.00%
  • VoiceBench In use Spoken instructions with speaker, environment and content variation. Voice assistant responses scored by GPT judge or rules. Qwen3.5-Omni-Plus 93.1%

Data audit

All 8 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
  • Pass Independent runs Scores from the official leaderboard (20 entries).
  • Pass Models scored 20 entries on its leaderboard, none a model we track.
  • Note Newest score No scores yet.
  • Note Frontier models No models scored yet.
  • Note Headroom No scores yet.
Leaderboard import 3 pass
  • Pass Importer We read the board on davischallenge.org: 20 entries at the last read, none a model we track.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check