84‐4: Fine‐grained Action Detection in Visual HCI with Spatial Mask Video Foundation Models
SID Symposium Digest of Technical Papers · 2026-06 · 57 (1) · 1171-1174
DOI: 10.1002/sdtp.70001_316 ↗Kurz gefasst
Fachartikel in der Zeitschrift SID Symposium Digest of Technical Papers, online erschienen am 11.09.2026. Verfasst von Haiyang Guo und 3 weiteren Autor:innen. Die Arbeit nennt 11 Literaturangaben.
Abstract (Originalsprache)
Visual human‐computer interaction (HCI) is a broad and pivotal research direction in display applications, where AI‐driven precise recognition of micro‐actions in videos enables more intelligent and natural interaction experiences for display devices. Conventional AI methods, such as using convolutional neural networks (CNNs) as feature encoders for action recognition, often suffer from limited accuracy and robustness across micro‐action and multi‐labels. To address these challenges, this paper introduces two innovations: 1) employing a video understanding large model as the foundational network for action detection, with self‐supervised pre‐training of the base model ; and 2) designing a tailored spatial mask module to precisely locate and identify actions of micro spatial variation. Experimental results on the MMAD52 dataset demonstrate that our algorithm achieves the state of the art performance of 52 microactions across seven categories, This technology can be applied in multiple visual HCI scenarios such as smart display devices and embodied AI perception systems.
Zitieren
APA
Guo, H., Song, L., Lv, B., & Jiang, X. (2026). 84‐4: Fine‐grained Action Detection in Visual HCI with Spatial Mask Video Foundation Models. SID Symposium Digest of Technical Papers, 57(1), 1171-1174. https://doi.org/10.1002/sdtp.70001_316
Vancouver
Guo H, Song L, Lv B, Jiang X. 84‐4: Fine‐grained Action Detection in Visual HCI with Spatial Mask Video Foundation Models. SID Symposium Digest of Technical Papers. 2026;57(1):1171-1174. doi:10.1002/sdtp.70001_316
BibTeX
@article{Guo2026,
author = {Guo, Haiyang and Song, Lei and Lv, Bin and Jiang, Xingqun},
title = {84‐4: Fine‐grained Action Detection in Visual HCI with Spatial Mask Video Foundation Models},
journal = {SID Symposium Digest of Technical Papers},
year = {2026},
volume = {57},
number = {1},
pages = {1171--1174},
doi = {10.1002/sdtp.70001_316}
}
EUVIMED