Referring Expression Segmentation
The task aims at labeling the pixels of an image or video that represent an object instance referred by a linguistic expression. In particular, the referring expression (RE) must allow the identification of an individual object in a discourse or scene (the referent). REs unambiguously identify the target instance.
Papers
Showing 1–10 of 145 papers
All datasetsRefCoCo valRefCOCO testARefer-YouTube-VOS (2021 public validation)RefCOCO+ test BA2D SentencesRefCOCOg-valJ-HMDBDAVIS 2017 (val)RefCOCOg-testRefCOCO testBPhraseCutRefCOCO
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | HINet | IoU overall | 0.68 | — | Unverified |
| 2 | RefVOS | IoU overall | 0.67 | — | Unverified |
| 3 | ClawCraneNet | IoU overall | 0.64 | — | Unverified |
| 4 | CMSA+CFSA | IoU overall | 0.62 | — | Unverified |
| 5 | RefVOS | IoU overall | 0.6 | — | Unverified |
| 6 | SgMg (Video-Swin-B) | AP | 0.59 | — | Unverified |
| 7 | SOC (Video-Swin-B) | AP | 0.57 | — | Unverified |
| 8 | ReferFormer (Video-Swin-B) | AP | 0.55 | — | Unverified |
| 9 | SOC (Video-Swin-T) | AP | 0.5 | — | Unverified |
| 10 | MANET | AP | 0.47 | — | Unverified |