CASTLE 2026 asks 185 multiple-choice questions over 600+ hours of synchronized multi-view egocentric video. We explore two approaches on top of a shared multimodal preprocessing layer, including per-person timelines, speaker-resolved transcripts, and multi-VLM caption ensembles. Approach A, SVA: Search-Verify-Answer, is a three-stage pipeline that hierarchically narrows to a primary window, verifies sub-windows with a VLM under four anti-confabulation rules, and fuses evidence with an LLM judge under an evidence-priority hierarchy. Approach B, TMKG: Temporal-Multimodal-Knowledge-Graph, is the contrast: it builds a temporal multimodal knowledge graph, locates a primary cell via graph search, and produces the final answer with a single grounded VLM. SVA reaches a leaderboard accuracy of 0.50 and is our final challenge submission; TMKG reaches 0.35.
@article{arxiv.2605.27800,
title = {CuriosAI Submission to the CASTLE Challenge at EgoVis 2026},
author = {Yuto Kanda and Hayato Tanoue and Takayuki Hori},
journal= {arXiv preprint arXiv:2605.27800},
year = {2026}
}
Comments
The 4th place solution for the CASTLE Challenge at the CVPR EgoVis Workshop 2026