Research map / Vision-language and multimodal / CLIP and zero-shot
Uni-Perceiver: Pre-Training Unified Architecture for Generic Perception for Zero-Shot and Few-Shot Tasks
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Prompt-tune with 1% downstream data instead of using the pretrained model without tuning | Uni-Perceiver · ImageNet-1k classification | Acc: 78.0 -> 80.2 | Helped |
| Prompt-tune with 1% downstream data rather than fine-tune with the same data | Uni-Perceiver · Flickr30k text retrieval | R@1: 78.4 -> 84.4 | Helped |
| Prompt-tune with 1% downstream data rather than fine-tune with the same data | Uni-Perceiver · COCO Caption CIDEr | CIDEr: 100.1 -> 109.6 | Mixed |
| Prompt-tune with 1% downstream data rather than fine-tune with the same data | Uni-Perceiver · MSVD video caption | B@4: 47.3 -> 54.8 | Helped |
| Prompt-tune with 1% downstream data rather than fine-tune with the same data | Uni-Perceiver · MSVD text retrieval | R@1: 49.6 -> 61.2 | Helped |
| Prompt-tune with 1% downstream data rather than fine-tune with the same data | Uni-Perceiver · VQA v2 test-dev Yes/No | score: 63.0 -> 63.0 | No clear effect |
| Prompt-tune with 1% downstream data rather than fine-tune with the same data | Uni-Perceiver · GLUE MNLI | Acc: 47.3 -> 60.1 | Helped |
| Fine-tune on 100% rather than 1% downstream data | Uni-Perceiver · ImageNet-1k classification | Acc: 80.2 -> 83.8 | Helped |
| Fine-tune on 100% rather than 1% downstream data | Uni-Perceiver · MSVD video caption | B@4: 54.8 -> 61.5 | Helped |
| Fine-tune on 100% rather than use zero-shot inference | Uni-Perceiver · VQA v2 test-dev Yes/No | score: 0.9 -> 84.8 | Helped |
Closest papers by meaning
- UniAP: Towards Universal Animal Perception in Vision via Few-Shot Learning (AAAI 2024 · similarity 0.89)
- Perception Encoder: The best visual embeddings are not at the output of the network (NeurIPS 2025 · similarity 0.85)
- Argus: A Compact and Versatile Foundation Model for Vision (CVPR 2025 · similarity 0.84)
- Meow-Omni 1: A Multimodal Large Language Model for Feline Ethology (NeurIPS 2026 · similarity 0.83)
- ViPER: Empowering the Self-Evolution of Visual Perception Abilities in Vision-Language Models (ICLR 2026 · similarity 0.83)
- FalconPerception-HD: High Density Perception via Reinforcement Learning (NeurIPS 2026 · similarity 0.83)
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.