Embedded Files
Call For Workshop Papers
We invite researchers and industry professionals to submit original papers related to the OEL-IP workshop at 2025 ICIP. Papers should be formatted according to the IEEE templates.
Keynote Speakers
Collaborative Perception: A Promising Solution for Practical Autonomous Driving
Cooperative perception is a critical paradigm for overcoming the inherent limitations of single-vehicle perception systems in autonomous driving, such as occlusion, limited sensing ranges, and adverse environmental conditions. By enabling real-time collaboration between vehicles and roadside units through vehicle-to-everything (V2X) communication, cooperative perception significantly enhances perception coverage, robustness, and accuracy, thereby advancing safety-critical decision-making. This keynote will introduce innovative methods to address key challenges in distributed cooperative perception, including spatio-temporal misalignment, communication constraints, and resource heterogeneity. The proposed solutions—spanning spatio-temporal feature fusion, dynamic feature alignment, efficient communication mechanisms, and vehicular edge computing—deliver substantial improvements in perception accuracy, computational efficiency, and scalability. This work lays a strong foundation for safer and more reliable autonomous driving systems, and the discussion will highlight future directions for strengthening multi-agent collaboration and overcoming the practical challenges of deploying cooperative perception.
Speaker: Prof. Sun is a leading researcher in AI-assisted intelligent transportation systems, the Internet of Vehicles, and mobile vehicular cloud/edge computing, with 40+ peer-reviewed publications, the Best Paper Award at IEEE GLOBECOM 2019, and a spotlighted contribution on sustainable wireless sensor networks across IEEE Transactions. He earned his Ph.D. in Electrical and Computer Engineering from the University of Ottawa, has led projects supported by Canada’s National Defense and NSERC, serves as a reviewer for IEEE Transactions on Wireless Communications and on TPCs for ACM DIVANet, IEEE ISCC, and IEEE ICC, and is currently at Duke Kunshan University.
Workshop Organizers
Liang Song
Fudan University
Kostas Plataniotis
University of Toronto
Victor C.M. Leung
SBMU, SZU, UBC
Jiangchuan Liu
Simon Fraser University
Yang Liu
Fudan University
Topics
Dynamic Environment Adaptation: Theoretical frameworks for stability-guaranteed OEL in visual systems.
Edge-Cloud Collaborative Learning: Distributed OEL architectures for resource-constrained devices and privacy-preserving federated visual intelligence.
Multi-modal Foundation Models: OEL-driven continual learning for vision-language models and cross-modal knowledge transfer.
Ethical and Transparent OEL: Addressing bias mitigation, fairness in adaptive systems, and creating open-source benchmarks for reproducible research.
Important Dates
28 May, 2025
Workshop Paper Submission
25 Jun, 2025
Workshop Paper Acceptance Notification
2 Jul, 2025
Workshop Camera-ready Paper Submission
16 Jul, 2025
Workshop Author Registration
About the OEL and OEL-IP Workshop
Online Evolutive Learning (OEL) is a groundbreaking paradigm in artificial intelligence, offering real-time adaptability and autonomous model optimization through multi-agent interaction and dynamic environmental adaptation. Unlike traditional machine learning approaches that struggle with shifting data distributions, OEL integrates principles from control theory, communication systems, and distributed intelligence to maintain robustness. This is especially critical in image processing, where rapid changes in lighting, sensor performance, and adversarial conditions demand flexible, evolving models.
The OEL and OEL-IP Workshop addresses three key challenges: the lack of adaptive mechanisms in standard image processing pipelines, the need for cross-disciplinary synergy among machine learning, control theory, and communication systems, and the growing demand for ethical, transparent model deployment. By bringing these diverse domains together, the workshop establishes standardized evaluation metrics, promotes effective collaboration, and offers practical guidance on responsible data usage.
Focusing on four core areas—dynamic environment adaptation, edge-cloud collaborative learning, multimodal foundation models, and ethical transparency—the workshop will feature keynote talks, panel discussions, a satellite image time-series adaptation competition, and a collaborative whitepaper on medical imaging. Industry leaders and academic experts will share best practices, explore new avenues for real-world applications, and discuss open challenges such as catastrophic forgetting and communication latency in distributed systems.
Building on the success of previous OEL events, this workshop will foster innovative dialogue, encourage international collaboration, and pave the way for next-generation adaptive visual intelligence systems. By unifying previously siloed research efforts, the OEL and OEL-IP Workshop aspires to establish OEL as a cornerstone methodology for future image processing solutions.
Page updated
Google Sites
Report abuse