A Lightweight Spatiotemporal Saliency Detection Framework for VR Panoramic Dynamic Scenes
Journal of Computing and Information Technology · 07.04.2026 · 34 (1) · 65
DOI: 10.20532/cit.2026.1006291 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Computing and Information Technology, online erschienen am 07.04.2026. Verfasst von Dezhi Kong und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Saliency detection in virtual reality (VR) panoramic dynamic scenes faces two major challenges: geometric distortion caused by equirectangular projection (ERP) and the high computational cost of modeling long-term temporal dependencies. To address these issues, we propose TAD-Net, a lightweight spatiotemporal saliency detection framework that integrates cubemap projection (CMP), temporal attention, knowledge distillation, and adversarial training. CMP efficiently reduces panoramic distortion while enabling standard 2D convolutional processing. A dual-stream network extracts spatial appearance and temporal motion features, and a temporal attention module enhances dynamic saliency discrimination. To reconcile the accuracy–latency trade-off, a heavy teacher model transfers long-range temporal knowledge to a lightweight student model via distillation, while adversarial training improves boundary sharpness. Extensive experiments on Salient360-Dynamic and VR-EyeDynamic demonstrate that TAD-Net achieves state-of-the-art performance, improving AUC-Judd by up to 5.2% while maintaining real-time inference at 35.1 FPS on an RTX 3080 GPU. Cross-dataset evaluation confirms robust generalization under domain shifts. The results indicate that the proposed projection–perception-distillation pipeline effectively balances geometric correction, temporal reasoning, and real-time constraints in VR applications.
Zitieren
APA
Kong, D., Hao, H., & Gao, B. (2026). A Lightweight Spatiotemporal Saliency Detection Framework for VR Panoramic Dynamic Scenes. Journal of Computing and Information Technology, 34(1), 65. https://doi.org/10.20532/cit.2026.1006291
Vancouver
Kong D, Hao H, Gao B. A Lightweight Spatiotemporal Saliency Detection Framework for VR Panoramic Dynamic Scenes. Journal of Computing and Information Technology. 2026;34(1):65. doi:10.20532/cit.2026.1006291
BibTeX
@article{Kong2026,
author = {Kong, Dezhi and Hao, Huijuan and Gao, Bo},
title = {A Lightweight Spatiotemporal Saliency Detection Framework for VR Panoramic Dynamic Scenes},
journal = {Journal of Computing and Information Technology},
year = {2026},
volume = {34},
number = {1},
pages = {65},
doi = {10.20532/cit.2026.1006291}
}
EUVIMED