Metraipolis › archive › dc_f01312bef117
TwinPast: Benchmarking Historical Evidence Use in Streaming Video Understanding with Evidence Contrastive Video Pairs — arXiv:2610.41983
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.41983
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T08:21:55.554Z
- SHA-256 of the archived text
3c226ab525a1bdb50fadea9279c82d16414e244fe2abbec72e9c5ab845704a23- Stable URL of this page
https://metraipolis.com/documento/dc_f01312bef117
How to cite
TwinPast: Benchmarking Historical Evidence Use in Streaming Video Understanding with Evidence Contrastive Video Pairs — arXiv:2610.41983 — arXiv (en), https://arxiv.org/abs/2610.41983 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f01312bef117 (Metraipolis archive dc_f01312bef117, sha256 3c226ab525a1bdb50fadea9279c82d16414e244fe2abbec72e9c5ab845704a23)
BibTeX
@misc{dcf01312bef117,
title = {TwinPast: Benchmarking Historical Evidence Use in Streaming Video Understanding with Evidence Contrastive Video Pairs — arXiv:2610.41983},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.41983},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f01312bef117; sha256 3c226ab525a1bdb50fadea9279c82d16414e244fe2abbec72e9c5ab845704a23},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41983. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
High accuracy on streaming video benchmarks does not establish whether a model uses visual history. Our review of OVO-Bench and StreamingBench finds that recent frames suffice for 32% of sampled historical questions, while models also answer many questions correctly from the question and options alone. We introduce TwinPast to test whether answers correctly follow changes in history, using 350 human-reviewed video pairs spanning five historical tasks. Each pair holds the recent video context, question, and ordered options fixed while changing critical historical evidence and the correct answer. Each video is evaluated independently, and PairAcc requires correct answers for both histories. We evaluate 34 model and input configurations. For Qwen3-VL-8B and DeepSeek-Flash, historical inputs improve performance on TwinPast, reversing the recent-frame advantage on existing benchmarks. The strongest evaluated streaming system reaches 49.14% PairAcc, compared with 66.57-91.43% for the evaluated general-purpose models. This gap persists across all five tasks. Case studies further show that models select states from the wrong time or contradict recorded event order despite retaining the relevant evidence. These findings highlight the need to connect historical memory and retrieval with reasoning that uses evidence according to the question's temporal references and event relations. Tema degli esploratori: statistica e scienza dei dati (elenco arXiv cs.CV, nuovi lavori del giorno). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2610.41983 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-11-03T08:06:38.851Z dalla pagina https://arxiv.org/list/cs.CV/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).