Open Vocabulary Object Detection
Open-vocabulary detection (OVD) aims to generalize beyond the limited number of base classes labeled during the training phase. The goal is to detect novel classes defined by an unbounded (open) vocabulary at inference.
Papers
Showing 1–10 of 145 papers
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | LaMI-DETR | AP novel-LVIS base training | 43.4 | — | Unverified |
| 2 | DITO | AP novel-LVIS base training | 40.4 | — | Unverified |
| 3 | OV-DQUO(ViT-L/14) | AP novel-LVIS base training | 39.3 | — | Unverified |
| 4 | CoDet (EVA02-L) | AP novel-LVIS base training | 37 | — | Unverified |
| 5 | CLIPSelf | AP novel-LVIS base training | 34.9 | — | Unverified |
| 6 | OVMR | AP novel-LVIS base training | 34.4 | — | Unverified |
| 7 | DE-ViT | AP novel-LVIS base training | 34.3 | — | Unverified |
| 8 | CFM-ViT | AP novel-LVIS base training | 33.9 | — | Unverified |
| 9 | CLIM (RN50x64) | AP novel-LVIS base training | 32.3 | — | Unverified |
| 10 | RO-ViT | AP novel-LVIS base training | 32.1 | — | Unverified |