7 papers
Fast and Accurate: An Adaptive VLA Inference Framework through Environment-aware Model Selection
Yuewei Sun, Lang Qin, Zechuan Tian +11
Embodied intelligence demands both long-horizon reasoning and real-time closed-loop responsiveness. Recent dual-system Vision-Language-Action (VLA) architectures combine fast react…
MAR: Efficient Large Language Models via Module-aware Architecture Refinement
Junhong Cai, Guiqin Wang, Kejie Zhao +6
Large Language Models (LLMs) excel across diverse domains but suffer from high energy costs due to quadratic attention and dense Feed-Forward Network (FFN) operations. To address t…
EdgeSync: Accelerating Edge-Model Updates for Data Drift through Adaptive Continuous Learning
Runchu Donga, Peng Zhao, Guiqin Wang +2
Real-time video analytics systems typically deploy lightweight models on edge devices to reduce latency. However, the distribution of data features may change over time due to vari…
Generative Model-Based Feature Attention Module for Video Action Analysis
Guiqin Wang, Peng Zhao, Cong Zhao +3
Video action analysis is a foundational technology within the realm of intelligent video comprehension, particularly concerning its application in Internet of Things(IoT). However,…
EdgeSync: Faster Edge-model Updating via Adaptive Continuous Learning for Video Data Drift
Peng Zhao, Runchu Dong, Guiqin Wang +1
Real-time video analytics systems typically place models with fewer weights on edge devices to reduce latency. The distribution of video content features may change over time for v…
Generative Model-based Feature Knowledge Distillation for Action Recognition
Guiqin Wang, Peng Zhao, Yanjiang Shi +2
Knowledge distillation (KD), a technique widely employed in computer vision, has emerged as a de facto standard for improving the performance of small neural networks. However, pre…
Weakly-Supervised Action Localization by Hierarchically-structured Latent Attention Modeling
Guiqin Wang, Peng Zhao, Cong Zhao +5
Weakly-supervised action localization aims to recognize and localize action instancese in untrimmed videos with only video-level labels. Most existing models rely on multiple insta…