Perception First: A Frontier Native-Video Model with Self-Consistency for Implicit Video Question Answering
Ali Alavi · 2026
Read Perception First: A Frontier Native-Video Model with Self-Consistency for Implicit Video Question Answering by Ali Alavi (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-01485.txt.
Description
Alavi reports a challenge entry built on a simple finding: perception comes first. The VRR Challenge asks multiple-choice questions about videos. Answers are not visible in any single frame. They must be inferred from layout, motion, and depth. Social context and causality also play a role. His team ran a training-free study of many video models. Inference tricks like chain-of-thought and decomposition were tested too.
Reasoning-side aids proved neutral to harmful in their study. Base model perception was the only reliable lever. Light test-time denoising also helped scores. Error analysis localizes difficulty to low-level perception. Depth, viewpoint, and counting are the hardest categories. Causal and social reasoning are nearly solved by contrast. A depth-cue prompt lowered test accuracy by 5 point 8 points.
Catalog record
| CREATOR | Ali Alavi |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 16 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.01485