Research map / Vision-language and multimodal / Human and embodied modeling
Improving Closed and Open-Vocabulary Attribute Prediction Using Transformers
Khoi Pham, Kushal Kafle, Zhe Lin, Zhihong Ding, Scott Cohen, Quan Tran, Abhinav Shrivastava
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 |
|---|---|---|---|
| pretrain on LSA then finetune on VAW (vs VAW only) | TAP · VAW | mAP: 65.4 -> 73.4 | Improved |
| use whole-image context with LSA pretraining (TAP vs RN50-Context) | TAP · VAW | mAP: 67.3 -> 73.4 | Improved |
| remove bounding boxes at test time | TAP · VAW | mAP: 73.4 -> 68.9 | Hurt |
| use CLIP text embeddings instead of ViCo (ablation) | OpenTAP · LSA common->rare | AP overall: 7.98 -> 10.91 | Improved |
| use CLIP text embeddings instead of BERT (ablation) | OpenTAP · LSA common->rare | AP overall: 9.90 -> 10.91 | Improved |
| use CLIP text embeddings instead of Phrase BERT (ablation) | OpenTAP · LSA common->rare | AP overall: 10.26 -> 10.91 | Improved |
| add Flickr30K-Entities to VG training data (ablation) | OpenTAP · LSA common->rare | AP overall: 7.04 -> 8.16 | Improved |
| add MS-COCO (ungrounded) to training data (ablation) | OpenTAP · LSA common->rare | AP overall: 8.16 -> 9.56 | Improved |
| add Localized Narratives (mouse trace) to training data (ablation) | OpenTAP · LSA common->rare | AP overall: 9.56 -> 10.91 | Improved |
| combine object-aware with object-agnostic prompts | CLIP · LSA common | AP overall: 2.64 -> 2.92 | Improved |
| replace CLIP with OpenTAP on unseen attributes | OpenTAP vs CLIP combined · LSA common->rare | AP unseen: 2.63 -> 5.37 | Improved |
| ensemble 1x, 1.25x, 1.5x crops of CLIP image embeddings | CLIP baselines · LSA | mAP: +0.4 mAP | Improved |
Closest papers by meaning
- OvarNet: Towards Open-Vocabulary Object Attribute Recognition (CVPR 2023 · similarity 0.90)
- Tree of Attributes Prompt Learning for Vision-Language Models (ICLR 2025 · similarity 0.88)
- Compositional Caching for Training-free Open-vocabulary Attribute Detection (CVPR 2025 · similarity 0.87)
- GUIDED: Granular Understanding via Identification, Detection, and Discrimination for Fine-Grained Open-Vocabulary Object Detection (NeurIPS 2025 · similarity 0.86)
- The Devil is in the Fine-Grained Details: Evaluating Open-Vocabulary Object Detectors for Fine-Grained Understanding (CVPR 2024 · similarity 0.85)
- DoGA: Enhancing Grounded Object Detection via Grouped Pre-Training with Attributes (AAAI 2025 · similarity 0.85)
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.
