Research map / Vision-language and multimodal / CLIP and zero-shot
LaFTer: Label-Free Tuning of Zero-shot Classifier using Language and Unlabeled Image Collections
Muhammad Jehanzeb Mirza, Leonid Karlinsky, Wei Lin, Horst Possegger, Mateusz Kozinski, Rogerio Feris, Horst Bischof
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 |
|---|---|---|---|
| apply LaFTer label-free tuning | CLIP ViT-B/32 · EuroSat | top-1 accuracy: 45.1 -> 73.9 | Improved |
| apply LaFTer label-free tuning | CLIP ViT-B/32 · ImageNet | top-1 accuracy: 61.9 -> 64.2 | Improved |
| use LaFTer instead of UPL | UPL · EuroSat | top-1 accuracy: 62.2 -> 73.9 | Improved |
| use LaFTer instead of UPL | UPL · DTD | top-1 accuracy: 48.0 -> 46.1 | Hurt |
| use LaFTer instead of UPL | UPL · ImageNet-A | top-1 accuracy: 26.9 -> 31.5 | Improved |
| remove learnable visual prompts (ablation) | LaFTer · 7-dataset average | top-1 accuracy: 61.4 -> 65.6 | Improved |
| add the classifier (instead of CLIP cosine similarity in both branches) (ablation) | LaFTer · 7-dataset average | top-1 accuracy: 60.4 -> 65.6 | Improved |
| add augmentations (ablation) | LaFTer · 7-dataset average | top-1 accuracy: 61.8 -> 65.6 | Improved |
| add affine parameter tuning (ablation) | LaFTer · 7-dataset average | top-1 accuracy: 62.0 -> 65.6 | Improved |
| add stop gradient (ablation) | LaFTer · 7-dataset average | top-1 accuracy: 64.2 -> 65.6 | Improved |
| add text-only pre-training (ablation) | LaFTer · 7-dataset average | top-1 accuracy: 63.7 -> 65.6 | Improved |
| add text-only pre-training (ablation) | LaFTer · EuroSat | top-1 accuracy: 69.7 -> 73.9 | Improved |
Later papers that compared with this method
2 later accepted papers report LafTer as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- SyncVL: Synchronizing Vision ⟷ Language Using Unsupervised Adaptation (ECCV 2026 · better on 1 of 1 shared result)
- Cascade Prompt Learning for Visual-Language Model Adaptation (ECCV 2024 · better on 1 of 1 shared result)
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
- Improved Zero-Shot Classification by Adapting VLMs with Text Descriptions (CVPR 2024 · similarity 0.90)
- Label Propagation for Zero-shot Classification with Vision-Language Models (CVPR 2024 · similarity 0.90)
- Large Language Models are Good Prompt Learners for Low-Shot Image Classification (CVPR 2024 · similarity 0.89)
- SPARC: Score Prompting and Adaptive Fusion for Zero-Shot Multi-Label Recognition in Vision-Language Models (CVPR 2025 · similarity 0.88)
- Few-Shot Recognition via Stage-Wise Retrieval-Augmented Finetuning (CVPR 2025 · similarity 0.88)
- Rethinking Few-Shot Adaptation of Vision-Language Models in Two Stages (CVPR 2025 · 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.
