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

Search research articles

Related Experiment Video

Updated: Oct 22, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

TaijiGNN: A New Cycle-Consistent Generative Neural Network for High-Quality Bidirectional Transformation between RGB

Xu Liu¹, Abdelouahed Gherbi¹, Wubin Li²

¹Synchromedia Laboratory, École de Technologie Supérieure (ÉTS), University of Québec, Montréal, QC H3C 1K3, Canada.

Sensors (Basel, Switzerland)

|August 28, 2021

Summary

Related Concept Videos

You might also read

Related Articles

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

Sort by

Same author

XGBoost-Based Digital Twin Model for Predicting Trajectory Errors in a Hexapod Coordinated Machining System Using Positioning Accuracy and Vibration Data.

Sensors (Basel, Switzerland)·2025

Same author

Enabling Fog-Blockchain Computing for Autonomous-Vehicle-Parking System: A Solution to Reinforce IoT-Cloud Platform for Future Smart Parking.

Sensors (Basel, Switzerland)·2022

Same author

Intuitive Development to Examine Collaborative IoT Supply Chain System Underlying Privacy and Security Levels and Perspective Powering through Proactive Blockchain.

Sensors (Basel, Switzerland)·2020

See all related articles

This summary is machine-generated.

The Taiji Generative Neural Network (TaijiGNN) effectively reconstructs multispectral images (MSIs) from RGB images by using dual generators and multilayer perceptron networks. This approach achieves state-of-the-art results with significantly less training data, overcoming challenges in spectral image translation.

Area of Science:

Computer Vision
Machine Learning
Image Processing

Background:

Reconstructing multispectral images (MSIs) from RGB images presents significant challenges due to differing information entropies and definitions.
Existing methods struggle with the spectrum transformation, particularly the underconstrained problem of MSI reconstruction from RGB data.

Purpose of the Study:

To propose a novel approach, the Taiji Generative Neural Network (TaijiGNN), for effective multispectral image reconstruction from RGB images.
To address the challenges of domain definition differences and information entropy imbalance between MSIs and RGBs.

Main Methods:

Developed TaijiGNN with two generators (G_MSI, G_RGB) forming dual cycles for bidirectional image translation.
Utilized multilayer perceptron (MLP) networks instead of traditional convolutional neural networks (CNNs) for generator implementation, enhancing simplicity and performance.

Keywords:

CycleGAN RGB image color vision computer vision cycle neural network hyperspectral image image processing image translation multilayer perceptron multispectral image spectral super-resolution

Related Experiment Videos

Last Updated: Oct 22, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

Modified loss functions by removing identity losses and incorporating consistent losses for paired images to improve training.

Main Results:

Achieved state-of-the-art performance on CAVE and ICVL datasets for spectral image translation.
Demonstrated the effectiveness of TaijiGNN with minimal training data, requiring only half the dataset for CAVE and one-fifth for ICVL to reach comparable results.
Showcased the system's ability to reach a dynamic balance through the complementary interaction of the two generators.

Conclusions:

TaijiGNN offers a robust and efficient solution for reconstructing multispectral images from RGB data.
The proposed MLP-based generators and modified loss functions significantly improve training effectiveness and reduce data requirements.
The dual-cycle architecture effectively handles domain translation challenges, paving the way for advanced spectral imaging applications.