Model card
EVA-02-L.
BAAI / PKUopen-sourceViT-L with EVA-CLIP pre-training (O365 + ImageNet-21K)
EVA-02 visual representation pretrained with masked image modeling. Large variant achieves 62.1 mask AP on LVIS v1.0 minival. Significantly improves rare category performance. CVPR 2023. arXiv:2303.11331.
§ 01 · Benchmarks
Every benchmark EVA-02-L has a recorded score for.
| # | Benchmark | Area · Task | Metric | Value | Rank | Date | Source |
|---|---|---|---|---|---|---|---|
| 01 | LVIS v1.0 | Computer Vision · Object Detection | mask-ap-rare | 61.0% | #2 | 2023-03-20 | source ↗ |
| 02 | LVIS v1.0 | Computer Vision · Object Detection | mask-ap | 62.1% | #4 | 2023-03-20 | source ↗ |
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 03 · Papers
1 paper with results for EVA-02-L.
- 2023-03-20· Computer Vision· 2 results
EVA-02: A Visual Representation Powerhouse for Dense Recognition Tasks
§ 05 · Sources & freshness
Where these numbers come from.
arxiv
2
results
2 of 2 rows marked verified.