Related Experiment Video
Updated: May 6, 2026
10:56
Long-term Behavioral Tracking of Freely Swimming Weakly Electric Fish
Published on: March 6, 2014
12.5K
A vision transformer based CNN for underwater image enhancement ViTClarityNet
Mohamed E Fathy1, Samer A Mohamed2,3, Mohammed I Awad2
1Mechatronics Engineering Department, Faculty of Engineering, Ain Shams University, Cairo, 11535, Egypt. 2002597@eng.asu.edu.eg.
Scientific Reports
|May 14, 2025
Summary
This study introduces ViT-Clarity, an advanced underwater image enhancement module using vision transformers and CNNs. It effectively improves underwater computer vision tasks like object detection, addressing challenges of poor visibility and scattering.
Area of Science:
- Computer Vision
- Image Processing
- Oceanography
Background:
- Underwater environments present significant challenges for computer vision due to light scattering, absorption, and poor illumination.
- Existing methods struggle to effectively enhance underwater images, limiting the performance of various vision tasks.
Purpose of the Study:
- To develop and evaluate an effective underwater image enhancement module, ViT-Clarity, that leverages vision transformers.
- To address the scarcity of paired underwater datasets by proposing a synthetic data generation method, BlueStyleGAN.
- To demonstrate the practical utility of enhanced underwater images in downstream computer vision applications.
Main Methods:
- Introduced ViT-Clarity, an underwater image enhancement module integrating vision transformers (ViT) with convolutional neural networks (CNN).
- Developed BlueStyleGAN, a generative adversarial network (GAN), to create realistic synthetic underwater images from clear in-air images.
- Evaluated ViT-ClarityNet on five diverse underwater datasets using quantitative (UCIQM, UCIQE, URanker) and qualitative metrics, comparing it against state-of-the-art methods and a transformer-free variant (ClarityNet).
Main Results:
- ViT-ClarityNet demonstrated superior performance in underwater image enhancement compared to existing methods and ClarityNet.
- BlueStyleGAN proved effective in generating realistic synthetic underwater images, showing good training stability.
- Enhanced images from ViT-Clarity significantly improved performance in object detection and SIFT feature matching tasks.
Conclusions:
- ViT-Clarity offers a robust solution for underwater image enhancement, significantly improving visibility and enabling more effective computer vision.
- The integration of vision transformers is crucial for achieving state-of-the-art performance in underwater image restoration.
- The proposed BlueStyleGAN addresses the data scarcity issue, facilitating further research and development in underwater vision.
Related Concept Videos
01:24Vision
48.7K
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.
48.7K
01:20Deconvolution
776
Deconvolution, also known as inverse filtering, is the process of extracting the impulse response from known input and output signals. This technique is vital in scenarios where the system's characteristics are unknown, and they must be inferred from the observable signals.
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
776
01:27Uniform Depth Channel Flow
890
Uniform depth channel flow keeps fluid depth consistent along channels such as irrigation canals. In natural channels, such as rivers, approximate uniform flow is often assumed. This condition occurs when the channel’s bottom slope matches the energy slope, balancing potential energy lost from gravity with head loss due to shear stress. This balance prevents depth changes along the channel length, resulting in a steady, uniform flow.Uniform flow in open channels with a constant...
890