CST-ViT: Cascaded Spatio–Temporal Redundancy Elimination for Efficient Vision Transformers on Edge IoT Devices

Transformer-based models have demonstrated outstanding performance in video understanding tasks due to their capacity to capture long-range dependencies. However, their high computational cost, along with the massive volume of streaming video data, presents significant challenges for real-time deployment on resource-constrained edge devices integrated into Internet of Things (IoT) systems. Existing approaches typically eliminate spatial or temporal redundancy in isolation, failing to fully exploit the inherent spatio–temporal similarity in video data. To address this limitation, we propose CST-ViT, a cascaded spatio–temporal redundancy elimination framework that jointly reduces dynamic temporal and intraframe spatial redundancy. CST-ViT incorporates three gating modules: 1) the direct temporal gate for matching unchanged backgrounds; 2) the offset temporal gate for capturing motion-related changes; and 3) the spatial gate for intraframe similarity matching. Together with a spatio–temporal caching and token reuse mechanism, CST-ViT enables efficient token filtering and computation reuse. Experimental results show that CST-ViT reduces computation by 55.88% with no loss in accuracy, and achieves up to a 74.75% reduction in computation with less than 1% accuracy degradation, outperforming state-of-the-art methods in terms of accuracy–efficiency tradeoff for video transformers.