Papers › LIGAR: Lightweight General-purpose Action Recognition

LIGAR: Lightweight General-purpose Action Recognition

30 Aug 2021arXiv:2108.13153archive 2025-07-28

Evgeny Izutov

Growing amount of different practical tasks in a video understanding problem has addressed the great challenge aiming to design an universal solution, which should be available for broad masses and suitable for the demanding edge-oriented inference. In this paper we are focused on designing a network architecture and a training pipeline to tackle the mentioned challenges. Our architecture takes the best from the previous ones and brings the ability to be successful not only in appearance-based action recognition tasks but in motion-based problems too. Furthermore, the induced label noise problem is formulated and Adaptive Clip Selection (ACS) framework is proposed to deal with it. Together it makes the LIGAR framework the general-purpose action recognition solution. We also have reported the extensive analysis on the general and gesture datasets to show the excellent trade-off between the performance and the accuracy in comparison to the state-of-the-art solutions. Training code is available at: https://github.com/openvinotoolkit/training_extensions. For the efficient edge-oriented inference all trained models can be exported into the OpenVINO format.

Paper PDF Code

Code

openvinotoolkit/training_extensions officialmentioned in paperpytorchApache-2.0 report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Action Recognition Gesture Recognition Hand-Gesture Recognition Video Understanding

Results from the paper archive 2025-07-28

Task	Dataset	Model	Metric	Value	Rank at snapshot	Leaderboard	Report
Action Recognition	Jester (Gesture Recognition)	X3D MobileNet-V3 LGD-GC	Val	95.56	#3 of 3	Archive leaderboard	report
Action Recognition	UCF101	X3D MobileNet-V3 LGD-GC	3-fold Accuracy	94.85	#50 of 91	Archive leaderboard	report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

1x1 Convolution 3D Convolution Average Pooling Batch Normalization CLIP Convolution Dense Connections Depthwise Convolution Depthwise Separable Convolution Dropout Global Average Pooling Hard Swish Inverted Residual Block Pointwise Convolution ReLU ReLU6 Sigmoid Activation Squeeze-and-Excitation Block

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections