ReVSI: Rebuilding Visual Spatial Intelligence Evaluation for Accurate Assessment of VLM 3D Reasoning

About this episode

The episode discusses the ReVSI benchmark aimed at improving the evaluation of visual spatial intelligence in vision-language models.

🤗 Upvotes: 57 | cs.CV Authors: Yiming Zhang, Jiacheng Chen, Jiaqi Tan, Yongsen Mao, Wenhu Chen, Angel X. Chang Title: ReVSI: Rebuilding Visual Spatial Intelligence Evaluation for Accurate Assessment of VLM 3D Reasoning Arxiv: http://arxiv.org/abs/2604.24300v1 Abstract: Current evaluations of spatial intelligence can be systematically invalid under modern vision-language model (VLM) settings. First, many benchmarks derive question-answer (QA) pairs from point-cloud-based 3D annotations originally curated for traditional 3D perception. When such annotations are treated as ground truth for video-based evaluation, reconstruction and annotation artifacts can miss objects that are clearly visible in the video, mislabel object identities, or corrupt geometry-dependent answers (e.g., size), yielding incorrect or ambiguous QA pairs. Second, evaluations often assume full-scene access, while many VLMs operate on sparsely sampled frames (e.g., 16-64), making many questions effectively unanswerable under the actual model inputs. We improve evaluation validity by introducing ReVSI, a benchmark and protocol that ensures each QA pair is answerable and correct under the model's actual inputs. To…

People in this episode

Hosts: Jingwen Liang, Gengyu Wang

Topics covered

• visual spatial intelligence
• evaluation
• 3D reasoning
• vision-language models
• benchmarking

Keywords

• ReVSI
• visual spatial intelligence
• 3D reasoning
• evaluation
• VLM
• benchmark
• QA pairs

Mentioned in this episode

Organizations: Arxiv

More episodes of Daily Paper Cast

• EvoArena: Tracking Memory Evolution for Robust LLM Agents in Dynamic Environments · June 13, 2026 · 24 min
• MiniMax Sparse Attention · June 13, 2026 · 26 min
• SpatialClaw: Rethinking Action Interface for Agentic Spatial Reasoning · June 13, 2026 · 23 min
• InterleaveThinker: Reinforcing Agentic Interleaved Generation · June 13, 2026 · 21 min
• FORT-Searcher: Synthesizing Shortcut-Resistant Search Tasks for Training Deep Search Agents · June 13, 2026 · 23 min
• Robust-U1: Can MLLMs Self-Recover Corrupted Visual Content for Robust Understanding? · June 13, 2026 · 20 min