[Challenge] Taiwan AI Safety Challenge (TAISC) in conjunction with IEEE AVSS 2025
Introduction
The Taiwan AI Safety Challenge invites researchers and practitioners to develop Vision–Language Model (VLM)–based solutions for future accident classification using a limited dashcam video dataset. Hosted on the CodeBench platform, the competition features a single track focused on Future Accident Classification, evaluated primarily by F1-Score, accuracy, and AUC. To support model development and fine-tuning on this limited data, additional aiDAPTIV+ computing resources will be available for eligible participants via an application process, allowing them to supplement their own capabilities. All submissions will undergo a quantitative, double‐blind evaluation.
Lead Organizers
Prof. Jun‑Wei Hsieh (National Yang Ming Chiao Tung University, Taiwan)
Prof. Chih‑Chung Hsu (National Yang Ming Chiao Tung University, Taiwan)
[Challenge] Natural language-based Automotive Video Identification (NAVI 2025)
Explore the intersection of language and vision at the NAVI 2025 Challenge, focused on Natural Language-based Automotive Video Identification. This challenge is hosted directly at AVSS 2025.
NAVI 2025 centres on leveraging real-world road surveillance videos paired with rich natural language descriptions to advance vehicle video retrieval. Participants will tackle tasks aiming to develop practical, large-scale applications using Computer Vision, Natural Language Processing, and Deep Learning. The ultimate goal is to enhance intelligent transportation systems and contribute towards building safer and smarter cities.
This challenge presents a unique opportunity for AVSS researchers interested in multi-modal analysis (vision + language) and practical AI applications in urban and transportation environments.
The challenge jointly organized by ETRI and UNIST.
[Challenge] Performance Evaluation of Tracking and Surveillance 2025 (PETS2025)
Introducing PETS2025, the 20th IEEE International Workshop on Performance Evaluation of Tracking and Surveillance. Held in conjunction with IEEE AVSS 2025 on August 27th, 2025, in Reading, UK, this workshop serves as a vital platform for researchers and practitioners within the AVSS community.
The core focus of PETS lies in the crucial task of evaluating the performance of computer vision algorithms for video surveillance, including areas like tracking, detection, and event recognition. It is particularly renowned for providing challenging public datasets and standardized evaluation methodologies, offering essential benchmarks for advancing the field.
If your work involves objectively evaluating and comparing the effectiveness of surveillance technologies, PETS2025 is a key event to follow.
[ E-mail ] Prof. James Ferryman (University of Reading, England)
[Challenge] AVSS 2025 Student Research Symposium & Awards Session
Introduction
This special session “AVSS 2025 Research Symposium & Awards Session” is hosted under AVSS 2025. Advanced visual and signal-based systems have shown remarkable success in various real-world applications. The applications of vision and signal are now shaping diverse domains such as automotive systems, drones, and security institutions.
This session provides a platform for Bachelor/PhD/Master students to enhance their engagement in the field of vision and signal processing. Graduate students are encouraged to submit research abstracts on any aspect of computer vision, signal processing, or related applications.
List of topics
Research topics include, but are not limited to, the following:
Signal Processing
Computer Vision
Multimedia
Vision-Language Models
Intelligent Transportation Systems
Artificial Intelligence
Machine Learning
Data Mining
Recommendation Systems
Classification
Clustering
Anomaly Detection
Robotics
AIoT
Embedded System
Submission Guidelines
This session is organized as a research competition. Participating Bachelor/PhD/Master students are required to submit a one-page extended abstract summarizing their original research contributions. The abstract should concisely present the motivation, methodology, and key findings of the work. Submissions will be reviewed by the session committee.
Authors of selected abstracts will be invited to the final round, where they will present their work as a competition oral presentation during the AVSS 2025 Graduate Research Symposium & Awards Session. Presentations will be evaluated by a panel of experts, and awards will be given to outstanding entries based on the quality, innovation, and impact of the work.
The submission website is hosted at https://cmt3.research.microsoft.com/AVSS2025/. One of the authors of each selected abstract must register and present the work at the session. An authorization agreement will be required following selection.
Best PhD student research 1st /2nd / 3rd award
Best Master student research 1st /2nd / 3rd award
Best Bachelor student research 1st /2nd / 3rd award
Best PhD/Master/ Bachelor Advisor award
Organizers
Jun-Wei Hsieh (National Yang-Ming Chiao-Tung University, Taiwan)
Ming-Ching Chang(State University of New York, USA)
Jenq-Neng Huang(University of Washington, USA)
Chih-Chung Hsu(National Yang-Ming Chiao-Tung University, Taiwan)
Chih-Fan Hsu(Inventec Corporation, Taiwan)
Yi-Zeng Hsieh (National Taiwan University of Science and Technology)
Yun-Hsin Chuang(National Cheng Kung University, Taiwan)
Guo-Shiang Lin(National Chin-Yi University of Technology, Taiwan)
Chuan-Wang Chang(National Chin-Yi University of Technology, Taiwan)