LiquidTAD: An Efficient Method for Temporal Action Detection via Liquid Neural Dynamics

2026-04-20

AI Summary

This work addresses the high computational complexity, parameter redundancy, and deployment challenges of existing Transformer-based temporal action detection methods by proposing LiquidTAD, the first framework to integrate parallelized liquid neural networks into this domain. At its core, the ActionLiquid module leverages the closed-form continuous-time (CfC) formulation to model temporal dynamics with linear complexity, efficiently capturing action dependencies while employing learnable time constants to adaptively modulate sensitivity to varying action durations. On THUMOS-14, LiquidTAD achieves a state-of-the-art average mAP of 69.46% with only 10.82M parameters—63% fewer than ActionFormer—and demonstrates superior accuracy-efficiency trade-offs and robustness to temporal sampling on both ActivityNet-1.3 and Ego4D benchmarks.

Technology Category

• Computer Vision: Video Understanding & Activity Analysis
• Machine Learning: Time-Series/Data Streams
• Planning, Routing, and Scheduling: Temporal Planning

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Temporal Action Detection (TAD) in untrimmed videos is currently dominated by Transformer-based architectures. While high-performing, their quadratic computational complexity and substantial parameter redundancy limit deployment in resource-constrained environments. In this paper, we propose LiquidTAD, a novel parameter-efficient framework that replaces cumbersome self-attention layers with parallelized ActionLiquid blocks. Unlike traditional Liquid Neural Networks (LNNs) that suffer from sequential execution bottlenecks, LiquidTAD leverages a closed-form continuous-time (CfC) formulation, allowing the model to be reformulated as a parallelizable operator while preserving the intrinsic physical prior of continuous-time dynamics. This architecture captures complex temporal dependencies with $O(N)$ linear complexity and adaptively modulates temporal sensitivity through learned time-constants ($τ$), providing a robust mechanism for handling varying action durations. To the best of our knowledge, this work is the first to introduce a parallelized LNN-based architecture to the TAD domain. Experimental results on the THUMOS-14 dataset demonstrate that LiquidTAD achieves a highly competitive Average mAP of 69.46\% with only 10.82M parameters -- a 63\% reduction compared to the ActionFormer baseline. Further evaluations on ActivityNet-1.3 and Ego4D benchmarks confirm that LiquidTAD achieves an optimal accuracy-efficiency trade-off and exhibits superior robustness to temporal sampling variations, advancing the Pareto frontier of modern TAD frameworks.

Problem

• Temporal Action Detection
• Computational Complexity
• Parameter Efficiency
• Resource-Constrained Deployment
• Untrimmed Videos

Innovation

• Liquid Neural Networks
• Temporal Action Detection
• Closed-form Continuous-time
• Parameter Efficiency
• Parallelizable Architecture