Guest Editors
Assist. Prof. Paolo Russo
Email: paolo.russo@diag.uniroma1.it
Affiliation: Department of Computer, Control and Management Engineering, Sapienza University of Rome, Rome, 00185, Italy
Homepage:
Research Interests: Deep Learning, Computer Vision, Monocular Depth Estimation, Signal Processing, Biomedical Classification
Assist. Prof. Fabiana Di Ciaccio
Email: fabiana.diciaccio@unifi.it
Affiliation: Department of Civil and Environmental Engineering, University of Florence, Via di Santa Marta, 3, 50139 Florence, Italy
Homepage:
Research Interests: Environmental Monitoring and machine/deep learning (sea surface temperature prediction, shoreline extraction, optimization of the orientation for maritime automated vehicles, etc); monitoring and preservation of cultural heritage against natural and anthropogenic risks and the effects of climate change, metrology, underwater photogrammetry and 3D reconstruction techniques.
Summary
In recent years, Vision Transformers (ViTs) have emerged as a powerful alternative to convolutional neural networks (CNNs) for computer vision tasks. However, their high computational cost and memory demands pose significant challenges for deployment in real-world applications, particularly in resource-constrained environments. Addressing these challenges, researchers have developed efficient ViT architectures, optimized training techniques, and novel performance benchmarking methodologies.
This special issue aims to gather state-of-the-art research on efficient Vision Transformers, covering algorithmic advancements, optimization techniques, and rigorous benchmarking. We invite contributions that explore the theoretical foundations, methodological innovations, and practical implementations of efficient ViTs. Papers demonstrating the versatility and scalability of these models across various vision tasks are particularly encouraged. By providing a platform for these advancements, this special issue seeks to promote collaboration and guide future research in making Vision Transformers more efficient and accessible.
The proposed special issue welcomes original research articles, surveys, and reviews on efficient Vision Transformers, including (but not limited to) the following topics:
· Development of lightweight and efficient Vision Transformer architectures
· Pruning, quantization, and low-rank approximations for ViTs
· Knowledge distillation techniques for compact Vision Transformers
· Hardware-aware and energy-efficient ViT models
· Efficient training strategies, including self-supervised and transfer learning for ViTs
· Performance benchmarking methodologies for efficient Vision Transformers
· Applications of efficient ViTs in real-world scenarios
· Integration of efficient ViTs with edge computing and IoT platforms
· Robustness, interpretability, and fairness in Vision Transformer models
· Challenges and solutions for large-scale training of Vision Transformers
We encourage submissions illustrating the impact of efficient ViTs across various domains, such as:
- Autonomous systems and robotics
- Medical imaging and health informatics
- Smart cities and intelligent transportation
- Augmented and virtual reality applications
- Industrial automation and manufacturing
- Surveillance and monitoring
By assembling cutting-edge research on efficient Vision Transformers, this special issue aims to serve as a valuable resource for researchers and practitioners, providing insights into current trends and future directions in this rapidly evolving field.
Keywords
Deep Learning, Neural Networks, Vision Transformers, Network Optimization
Published Papers
Login
Register
Submit a Paper
Propose a Special lssue