Edge‑Cloud AI Powers Student Fatigue‑Driving Detection – Challenge Cup Winners

18th "Challenge Cup" National College Student Extracurricular Academic Science Competition – "Challenge the Leader" Special Event

Since February, 21 enterprises released cutting‑edge, application‑oriented topics; nearly 2,000 projects were submitted, setting a new record.

Huawei Cloud launched the “Edge‑Cloud Collaborative Fatigue Driving Intelligent Recognition” topic, attracting about 1,000 participants. On October 29, five top‑award teams demonstrated their solutions at Guizhou University.

Champion Project: Guizhou University – One‑Stage Target Detection Algorithm for Fatigue Driving

Supervisors: Cui Yunhe, Shen Guowei, Wang Lihui. Team leader: Fu Yujie.

The “ZhiYun Assistant” collects abnormal data during detection, performs incremental training to improve model accuracy, and leverages idle computing resources from nearby HarmonyOS vehicles for edge‑cloud collaboration. When fatigue is detected, it issues voice alerts and sends SMS to relatives.

Best Popularity Award: Huazhong University of Science and Technology – SmartEye: Human‑Behavior Analysis System for Fatigue Driving

Supervisors: Xiao Yang, Cao Zhiguo, Zhu Yingying. Team leader: Zeng Wenzheng.

The solution offers accuracy, stability, flexibility, and speed. It introduces a two‑stage anomaly detection algorithm, a light‑adaptive enhancement for low‑light conditions, and a few‑shot incremental learning method, enabling operation on low‑power devices and complex environments.

First Prize: Nanchang Hangkong University – Target‑Detection‑Based Fatigue Driving Recognition

Supervisors: Chen Ying, Wu Bo. Team leader: Zhang Wei.

The team developed four core algorithms: data‑iteration expansion, focused‑category loss, key‑frame rapid localization, and edge‑cloud collaborative strategy, reducing labeling effort, improving prediction accuracy, and shortening detection time.

First Prize: Anhui University – CICG Fatigue Driving Recognition for Primary Students

Supervisors: Chen Jie, Huang Zhixiang, Li Yingsong. Team leader: Deng Yingjian.

The model combines data migration and knowledge distillation for a lightweight high‑precision AI, includes a dedicated eye‑closure detector, and is optimized for extreme night‑time driving scenarios.

First Prize: Zhejiang Normal University – Zhijia Geek Dream‑Team: Edge‑Cloud Collaborative Fatigue and Distraction Detection System

Supervisors: Xu Huiying, Zhu Xinzong, Zhu Zhecheng. Team leader: Li Chen.

The system uses edge devices and cloud platforms for real‑time detection and warning of fatigue or distraction, achieving high precision, low false‑alarm rate, lightweight deployment, fast response, and low resource consumption.

Huawei Cloud employed ModelArts automatic scoring, using test‑set videos for batch inference and evaluation, allowing teams to view daily scores and ensuring fair competition. Experts provided topic‑interpretation training and offered computing resources and technical support.