Research map / Vision-language and multimodal / CLIP and zero-shot
ORION: ORthonormal Text Encoding for Universal VLM AdaptatION
Omprakash Chakraborty, Jose Dolz, Ismail Ben Ayed
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add ORION text encoder | CLIP ViT-B/16 · Zero-shot, 11 datasets | Top-1 accuracy: 63.70 -> 66.46 | Improved |
| add ORION text encoder | CLIP ViT-L/14 · Zero-shot, 11 datasets | Top-1 accuracy: 71.34 -> 72.85 | Improved |
| add ORION text encoder | MetaCLIP · Zero-shot, 11 datasets | Top-1 accuracy: 69.07 -> 69.80 | Improved |
| add ORION initialization | CoOp · Few-shot 1-shot, DTD | Top-1 accuracy: +7.6 | Improved |
| add ORION initialization | CLAP · Few-shot 4-shot, DTD | Top-1 accuracy: +13.6 | Improved |
| add ORION initialization | CLAP · Few-shot 16-shot, 11 datasets | Top-1 accuracy: 70.04 -> 74.99 | Improved |
| add ORION initialization | CoOp · Few-shot 16-shot, ImageNet | Top-1 accuracy: 71.9 -> 71.8 | No clear effect |
| add ORION text encoder | MTA · Test-time adaptation, 11 datasets | Top-1 accuracy: 65.87 -> 67.53 | Improved |
| add ORION text encoder | TPT · Test-time adaptation, 11 datasets | Top-1 accuracy: 65.09 -> 66.45 | Improved |
| add ORION text encoder | StatA · Batch-realistic batch 64, Very Low | Top-1 accuracy: 70.35 -> 71.21 | Improved |
| add ORION text encoder | StatA · Online-realistic Separate | Top-1 accuracy: 69.06 -> 70.53 | Improved |
| add ORION text encoder | CuPL · EuroSAT | Top-1 accuracy: 56.6 -> 58.67 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CLAP — CLAP: Isolating Content from Style through Contrastive Learning with Augmented Prompts (ECCV 2024) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Frustratingly Easy Test-Time Adaptation of Vision-Language Models (NeurIPS 2024 · similarity 0.90)
- Language Models as Black-Box Optimizers for Vision-Language Models (CVPR 2024 · similarity 0.89)
- VladVA: Discriminative Fine-tuning of LVLMs (CVPR 2025 · similarity 0.89)
- Realistic Test-Time Adaptation of Vision-Language Models (CVPR 2025 · similarity 0.89)
- VLM2Vec: Training Vision-Language Models for Massive Multimodal Embedding Tasks (ICLR 2025 · similarity 0.89)
- A Closer Look at the Few-Shot Adaptation of Large Vision-Language Models (CVPR 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
