Research map / 3D vision / 3D generation
Beyond Single-View Sufficiency: CVBench for Cross-View Human Understanding
Tianchen Guo, Chen Liu, Xin Yu
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 |
|---|---|---|---|
| scale open-source MLLM from 7B to 72B/78B | Qwen2.5-VL · CVBench spatial (All) | accuracy (%): 27.2 -> 33.3 | Improved |
| use closed-source model instead of best open-source model | MLLMs · CVBench spatial (All) | accuracy (%): 34.8 -> 38.6 | Improved |
| use closed-source model instead of best open-source model | MLLMs · CVBench temporal (All) | accuracy (%): 32.8 -> 34.9 | Improved |
| move from coarse-grained to fine-grained tasks | GPT-5 · CVBench spatial | accuracy (%): 41.2 -> 35.5 | Hurt |
| move from spatial to temporal tasks | InternVL3-78B · CVBench | accuracy (%): 34.8 -> 32.8 | Hurt |
| remove images (text-only Blind baseline) (ablation) | Gemini-2.5-Flash · CVBench spatial (All) | accuracy (%): 37.1 -> 18.5 | Hurt |
| compare best model to human performance | Gemini-2.5-Pro · CVBench spatial (All) | accuracy (%): 94.4 -> 38.6 | Hurt |
Closest papers by meaning
- Seeing from Another Perspective: Evaluating Multi-View Understanding in MLLMs (AAAI 2026 · similarity 0.89)
- Human-MME: A Holistic Evaluation Benchmark for Human-Centric Multimodal Large Language Models (ICLR 2026 · similarity 0.88)
- Eyes on VLM: Benchmarking Gaze Following and Social Gaze Prediction in Vision Language Models (NeurIPS 2026 · similarity 0.88)
- MMT-Bench: A Comprehensive Multimodal Benchmark for Evaluating Large Vision-Language Models Towards Multitask AGI (ICML 2024 · similarity 0.87)
- JourneyBench: A Challenging One-Stop Vision-Language Understanding Benchmark of Generated Images (NeurIPS 2024 · similarity 0.87)
- HumanPCR: Probing MLLM Capabilities in Diverse Human-Centric Scenes (ICLR 2026 · similarity 0.87)
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.
