Metraipolis › archive › dc_1e2dd6fa4dd9

Disentangling Where and When:Factored Spatio-Temporal Explanations for Video Action Recognition — arXiv:2611.11214

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.11214
License (as archived)
CC0-1.0
Archived
2026-11-10T18:43:10.609Z
SHA-256 of the archived text
6d3100af84b7d65f107d24fe43e189648b155fad1afa39c5198889bf8f0356c8
Stable URL of this page
https://metraipolis.com/documento/dc_1e2dd6fa4dd9

How to cite

Disentangling Where and When:Factored Spatio-Temporal Explanations for Video Action Recognition — arXiv:2611.11214 — arXiv (en), https://arxiv.org/abs/2611.11214 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_1e2dd6fa4dd9 (Metraipolis archive dc_1e2dd6fa4dd9, sha256 6d3100af84b7d65f107d24fe43e189648b155fad1afa39c5198889bf8f0356c8)
BibTeX
@misc{dc1e2dd6fa4dd9,
  title = {Disentangling Where and When:Factored Spatio-Temporal Explanations for Video Action Recognition — arXiv:2611.11214},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.11214},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_1e2dd6fa4dd9; sha256 6d3100af84b7d65f107d24fe43e189648b155fad1afa39c5198889bf8f0356c8},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.11214. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Post-hoc explanation methods for video action recognition typically conflate spatial appearance and temporal dynamics into a single saliency or importance map. We propose a perturbation-based framework that disentangles these contributions into a spatial appearance mask ($M_a$) and a temporal motion mask ($M_m$). The two masks are jointly optimized against a frozen action recognition model through factored perturbation architecture, with nearest frame replacement to identify important temporal dynamics. A separation penalty discourages the appearance mask from spending budget on frames that the motion mask has already discarded. A ReLU-based area constraint penalizes explanations larger than an assigned budget. To evaluate these explanations, we introduce a separation protocol that sweeps spatial pixels and temporal frames independently, backed by proofs characterizing the proposed method. Across EPIC-Kitchens-55, EGTEA Gaze+, and Something-Something V2, our method attains the lowest deletion AUC against leading baselines while producing masks roughly five times sparser.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2611.11214 — 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 2026-11-10T18:37:37.628Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).