
Motivation
The increasing integration of Artificial Intelligence (AI) into diverse applications calls for innovative approaches that emphasize efficiency, sustainability, and scalability to reduce the associated environmental costs (Green AI) while extending the range of devices capable of running this kind of technology. This issue is particularly evident in the domains of Internet of Things (IoT) and Computer Vision systems, where computational and energy constraints demand tailored solutions.The increasing complexity of AI systems and models, with their growing numbers of parameters and hyperparameters, has driven exponential increases in resource requirements for training and deployment.
On the other hand, many applications have become too computationallydemanding to be executed on devices with limited processing power, creating a significant barrier to widespread adoption.The concept of Green AI provides a framework for addressing these challenges. It includes strategies for designing energy-efficient Machine Learning (ML) and Deep Learning (DL) models (green-in AI) and leveraging AI to promote sustainable practices in other fields (green-by AI).These efforts are supported by tools that measure and optimise energy consumption, alongside explainable AI (XAI) techniques to better understand and enhance system efficiency.
Lightweight models can enable these technologies to operate effectively in resource constrained environments while minimizing computational costs. The size and inference speed of AI models are often overlooked when proposing solutions that rely on ensembles of models with millions of parameters. Such approaches are impractical for systems that need to process vast amounts of images and videos daily, as seen in applicationslike social network content analysis (e.g., detecting fake images and videos) or real-time scenarios requiring immediate responses.This workshop aims to bring together researchers and practitioners to exchange insights, strategies, and tools for advancing environmentally sustainable and efficient AI technologies.
The focus spans the design and deployment of lightweight models capable of addressing computational and energy constraints in resource-constrained domains, such as IoT and Computer Vision systems.
List of relevant topic areas
-
Energy-Efficient AI Models for IoT and Edge Devices.
-
Lightweight Machine/Deep Learning Models for Real-Time and Computer Vision Applications.
-
Lightweight Machine/Deep Learning Models for Resource-Constrained Devices.
-
Low-Resource AI for Computer Vision-Based Applications.
-
Low-Resource AI for IoT and Embedded Systems.
-
Optimization of Machine/Deep Learning Models for Energy and Computational Efficiency.
-
Scalable AI Architectures for Low-Power Devices and Embedded Systems
-
AI for Sustainable Practices in IoT Systems.
-
Green-by-AI: AI Applications to Promote Sustainability in Other Domains (i.e., smart agriculture, healthcare, etc.).
-
AI for Real-Time Applications and Optimisation of Resource Usage.
-
Explainable AI (XAI) Techniques for Enhancing Efficiency and Transparency.
-
Energy-Efficient Machine/Deep Learning in Mobile and Wearable Devices.
-
AI Solutions for Deepfake Detection with Low Computational Overhead.
-
Lightweight AI Models for Real-Time Deepfake Detection.
-
AI-Based Security Systems with Low Computational Overhead.
-
Lightweight Biometrics Recognition Systems for Secure Authentication.
-
AI-Driven Solutions for Enhancing Energy Efficiency in Distributed Networks.
-
Privacy-Preserving AI Models for Sustainable Data Processing.
Important Deadlines
The key deadlines for paper submission are aligned with those of the main conference.
SUBMISSION
SITE OPENING
ABSTRACT
DEADLINE
PAPER
DEADLINE
REBUTTAL
PERIOD
AUTHOR
NOTIFICATION
10 April 2025
29 April 2025
6 May 2025
23-25 June 2025
10 July 2025
Special Issues and Fast Track Opportunity
The workshop will not publish official proceedings; however, all authors of accepted papers will be invited to revise and extend their work for submission to one of the following four venues, ensuring that the submitted version aligns with the scope and objectives of the selected venue:
​
-
Multimedia Tools and Applications: "Explainable AI (XAI) for Biometric Authentication and Medical Imaging: A Cross-Disciplinary Challenge"
-
Image and Vision Computing: "Security-AI: Attacks on AI Systems in Computer Vision"
-
International Journal of Educational Technology in Higher Education:"Harnessing Large Language Models for Teaching and Learning: Challenges, Opportunities, and Future Directions"
-
Image and Vision Computing (Fast Track): regular submission/full length article (regular paper) to Image and Vision Computing, benefiting from an accelerated review process.
​​
The final decision on the acceptance of those papers will be always taken by the Editors-in-Chief of each journals.
Registration and workshop attending
Authors of accepted papers, or at least one of them, are requested to register to the ECAI-2025 Registration page (available later) and present their work at the conference, otherwise their papers will not be considered for journals pubblication.
Instructions For Authors
Submissions must use the LaTeX template provided at this link and should not exceed 8 pages in length (including all figures, tables, and references).
​
Submissions are made through the Easychair website (https://easychair.org/conferences?conf=sails2025).
Organization Team

Prof. Aniello Castiglione

Dr. Matteo Polsinelli

Dr. Lucia Cimmino
(Lead Contact)

Prof. Xinggang Wang

Prof. Danilo Mandic