Research map / Vision-language and multimodal / CLIP and zero-shot
b-CLIP: Text-Conditioned Contrastive Learning for Multi-Granular Vision-Language Alignment
Fatimah Zohra, Chen Zhao, Hani Itani, Bernard Ghanem
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 text-conditioned multi-granular alignment (b-CAL, K=6, CE) versus plain global-loss fine-tuning (ablation) | CLIP ViT-B/16 · FG-OVD Hard | accuracy: 22.0 -> 29.2 | Improved |
| add multiple sentence positives with plain global loss (no text-conditioning) (ablation) | CLIP fine-tuning · FG-OVD Hard | accuracy: 22.0 -> 19.7 | Hurt |
| use beta > 0 instead of strict self-matching (CE) (ablation) | b-CLIP CE K=36 · FG-OVD Hard | accuracy: 3.6 -> 30.9 | Improved |
| use beta > 0 instead of strict self-matching (BCE) (ablation) | b-CLIP BCE K=36 · FG-OVD Hard | accuracy: 14.8 -> 20.2 | Improved |
| raise beta from 0.5 to 1 (CE) (ablation) | b-CLIP CE K=36 · FG-OVD Hard | accuracy: 30.9 -> 30.8 | No clear effect |
| increase hierarchy size K from 6 to 36 (CE) (ablation) | b-CLIP CE · FG-OVD Hard | accuracy: 29.2 -> 30.9 | Improved |
| increase hierarchy size K from 6 to 36 (BCE) (ablation) | b-CLIP BCE · FG-OVD Medium/Easy | accuracy: Easy 45.3 -> 34.2 | Hurt |
| use BCE instead of CE loss (ablation) | b-CLIP · Urban-1k T2I | R@1: 89.0 -> 91.8 | Improved |
| use BCE instead of CE loss (ablation) | b-CLIP · FG-OVD Hard | accuracy: 30.9 -> 20.1 | Hurt |
| use BCE variant for short-caption retrieval (ablation) | b-CLIP · MSCOCO T2I | R@1: 34.0 -> 39.8 | Improved |
| infer with text-conditioned image embeddings instead of CLS (CE, K=6) (ablation) | b-CLIP CE · Urban-1k T2I | R@1: 87.9 -> 99.1 | Improved |
| infer with text-conditioned image embeddings at K=36 (BCE) (ablation) | b-CLIP BCE · Urban-1k T2I | R@1: 94.4 -> 88.9 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- Smart-CLIP — SmartCLIP: Modular Vision-language Alignment with Identification Guarantees (CVPR 2025) · better on 3 of 3 shared results
- Long-CLIP — Long-CLIP: Unlocking the Long-Text Capability of CLIP (ECCV 2024) · better on 3 of 3 shared results
- Fine-CLIP — FineCLIP: Self-distilled Region-based CLIP for Better Fine-grained Understanding (NeurIPS 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
- SmartCLIP: Modular Vision-language Alignment with Identification Guarantees (CVPR 2025 · similarity 0.91)
- GOAL: Global-local Object Alignment Learning (CVPR 2025 · similarity 0.90)
- Mining Fine-Grained Image-Text Alignment for Zero-Shot Captioning via Text-Only Training (AAAI 2024 · similarity 0.90)
- FineLIP: Extending CLIP's Reach via Fine-Grained Alignment with Longer Text Inputs (CVPR 2025 · similarity 0.90)
- Post-pre-training for Modality Alignment in Vision-Language Foundation Models (CVPR 2025 · similarity 0.90)
- SuperCLIP: CLIP with Simple Classification Supervision (NeurIPS 2025 · similarity 0.90)
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.
