Abstract:This technical report presents our solution, EgoAdapt (Egocentric Adaptation via Category, Calibration, and Consistency), to the CVPR 2026 HD-EPIC VQA challenge. HD-EPIC evaluates whether a vision-language model can reason over realistic first-person kitchen videos, where the evidence for an answer may be a short hand-object interaction, a long recipe trajectory, a spatial relation to a fixture, or a subtle gaze cue. The benchmark contains 26K multiple-choice questions across seven macro-categories: recipe, ingredient, nutrition, fine-grained action, 3D perception, object motion, and gaze. We observe that the main difficulty is not only model capacity, but also the mismatch between a single generic inference recipe and the heterogeneous temporal, spatial, and semantic structure of the benchmark. Our method, EgoAdapt, introduces three inference-time components: (1) category-conditioned routing with per-category prompts, frame budgets, and sampling rates; (2) calibrated option scoring that evaluates all candidate answers with letter-token likelihoods and generation agreement instead of relying only on direct generation; and (3) test-time consistency adaptation that aggregates predictions across option permutations and verification-style prompts for ambiguous cases. This design substantially improves over the available HD-EPIC baselines.
| Comments: | Technical Report for CVPR 2026 HD-EPIC VQA Challenge |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2605.24500 [cs.CV] |
| (or arXiv:2605.24500v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2605.24500 arXiv-issued DOI via DataCite (pending registration) |
From: Zhiwei Chen [view email]
[v1]
Sat, 23 May 2026 10:16:51 UTC (463 KB)
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。