LiquidTAD: Efficient Temporal Action Detection via Parallel Liquid-Inspired Temporal Relaxation

TL;DR

LiquidTAD introduces a hardware-efficient, parallel liquid-inspired relaxation method for temporal action detection, achieving competitive accuracy with significantly fewer parameters and lower computational costs.

Contribution

It proposes a parallel, vectorized relaxation mechanism inspired by liquid neural dynamics, enabling hardware-agnostic, linear-complexity temporal action detection.

Findings

Achieves 69.46% mAP on THUMOS-14 with 10.82M parameters.

Reduces parameter count by over 60% compared to ActionFormer.

Maintains competitive accuracy while lowering model footprint.

Abstract

Temporal Action Detection (TAD) requires precise localization of action boundaries within long, untrimmed video sequences. While current high-performing methods achieve strong accuracy, they are often characterized by excessive parameter counts, substantial computational overhead, and a reliance on specialized operators that hinder deployment across diverse hardware platforms. This paper presents LiquidTAD, a framework that distills the exponential relaxation prior of liquid neural dynamics into a parallel temporal operator, rather than reproducing full Liquid Neural Network (LNN) dynamics. By introducing a Parallel Liquid-inspired Relaxation mechanism, sequential ODE solving is avoided through a fully vectorized, non-recursive formulation built entirely upon standard neural operations, enabling hardware-agnostic deployment with linear complexity with respect to the temporal length. A complementary Hierarchical Decay-Rate Sharing Strategy further adapts this relaxation prior across feature pyramid levels, stabilizing optimization and implicitly compensating for temporal compression in deeper layers. Experimental evaluations on THUMOS-14 and ActivityNet-1.3 demonstrate that LiquidTAD achieves accuracy competitive with strong baselines while substantially lowering the model footprint. Specifically, on THUMOS-14, LiquidTAD achieves 69.46\% average mAP with only 10.82M parameters and 27.17G FLOPs, reducing the parameter count by over 60\% compared with ActionFormer.