Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Related Concept Videos

Vision

Vision

Vision is the result of light being detected and transduced into neural signals by the retina of the eye. This information is then further analyzed and interpreted by the brain. First, light enters the front of the eye and is focused by the cornea and lens onto the retina—a thin sheet of neural tissue lining the back of the eye. Because of refraction through the convex lens of the eye, images are projected onto the retina upside-down and reversed.

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Multiaxial Fatigue Life Assessment of Large Welded Flange Shafts: A Continuum Damage Mechanics Approach.

Materials (Basel, Switzerland)·2025

Same author

GR-AttNet: Robotic grasping with lightweight spatial attention mechanism.

PloS one·2025

Same author

Dual-branch differential channel hypergraph convolutional network for human skeleton based action recognition.

PloS one·2025

Same author

MEP-YOLOv5s: Small-Target Detection Model for Unmanned Aerial Vehicle-Captured Images.

Sensors (Basel, Switzerland)·2025

Same author

Highly active and reversible NiPSe<sub>3</sub> anode for sodium-ion batteries: enabling ultrafast sodium storage with exceptional cycling stability.

Journal of colloid and interface science·2025

Same author

Use of BOIvy Optimization Algorithm-Based Machine Learning Models in Predicting the Compressive Strength of Bentonite Plastic Concrete.

Materials (Basel, Switzerland)·2025

Search research articles

Related Experiment Video

Updated: Jul 25, 2026

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

LS-VIT: Vision Transformer for action recognition based on long and short-term temporal difference.

Dong Chen^1,2,3, Peisong Wu^1,3, Mingdong Chen^1,3

¹College of Physics and Electronic Engineering, Nanning Normal University, Nanning, China.

Frontiers in Neurorobotics

|November 15, 2024

Summary

This study introduces the Long and Short-term Temporal Difference Vision Transformer (LS-VIT) for efficient 3D video action recognition. The LS-VIT model achieves high accuracy by effectively capturing both short-term and long-term motion details in videos.

Keywords:

Vision Transformer action recognition deep learning motion extraction temporal crossing fusion

More Related Videos

Author Spotlight: Revolutionizing Remote Surgery with Augmented Reality and Robotics for Enhanced Precision and Accessibility

Author Spotlight: Revolutionizing Remote Surgery with Augmented Reality and Robotics for Enhanced Precision and Accessibility

Published on: August 9, 2024

Author Spotlight: Deciphering Electrical Networks Behind Complex Brain Activities and Disorders

Author Spotlight: Deciphering Electrical Networks Behind Complex Brain Activities and Disorders

Published on: November 1, 2024

Related Experiment Videos

Last Updated: Jul 25, 2026

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

Author Spotlight: Revolutionizing Remote Surgery with Augmented Reality and Robotics for Enhanced Precision and Accessibility

Author Spotlight: Revolutionizing Remote Surgery with Augmented Reality and Robotics for Enhanced Precision and Accessibility

Published on: August 9, 2024

Author Spotlight: Deciphering Electrical Networks Behind Complex Brain Activities and Disorders

Author Spotlight: Deciphering Electrical Networks Behind Complex Brain Activities and Disorders

Published on: November 1, 2024

Area of Science:

Computer Vision
Artificial Intelligence
Machine Learning

Background:

Transformer models excel in 2D vision but face computational challenges in 3D video tasks like action recognition.
Directly applying temporal transformations to 3D video data increases computational and memory demands due to data patch multiplication and complex self-attention mechanisms.

Purpose of the Study:

To develop an efficient and precise 3D self-attentive model for video action recognition.
To address the computational challenges posed by transformer models in 3D video analysis.

Main Methods:

Introduced the Long and Short-term Temporal Difference Vision Transformer (LS-VIT).
Incorporated short-term motion details by weighting differences across consecutive frames.
Integrated a module for long-term motion understanding using motion excitation and temporal differences from various segments.

Main Results:

LS-VIT achieved high recognition accuracy on multiple benchmarks, including UCF101, HMDB51, and Kinetics-400.
The model effectively models both short-term and long-term motion dynamics in videos.

Conclusions:

LS-VIT demonstrates strong performance in 3D video action recognition.
The model shows potential for further optimization to enhance real-time performance and action prediction capabilities.