The China Graduate Electronic Design Competition, guided by the Department of Degree Management and Graduate Education of the Ministry of Education, jointly organized by the China Association for Science and Technology Youth Science and Technology Center and the China Electronics Society, is one of the thematic events under the 'China Graduate Innovation Practice Series Contest.' It is a team-based electronic design innovation and creative practice activity for graduate students nationwide, actively participated in by universities and research institutes across the country. This competition, known as the 'Research Electronic Competition,' is oriented toward industry demands and has become a mainstream event that fosters innovation and entrepreneurship consciousness among graduate students in electronic-related disciplines, enhances teamwork, boosts practical abilities, and promotes the comprehensive growth of young innovative entrepreneurs.
The competition is characterized by its high standard, deep participation, high comprehensive quality requirements, and close integration with the industry. It consists of two major parts: technical competitions and a special section for business proposals, offering both self-proposed and corporate-themed categories. AlgoTech has joined through the corporate-themed category and established special awards, focusing on the design of edge computing systems based on TPU processors. AlgoTech also provides participating contestants with powerful development boards, drones, deep learning cars, and other hardware resources.
The 18th China Graduate Electronic Design Contest Grand Finals——SOPHGO Challenge Award List | |||
---|---|---|---|
Prizes | Team Name | College | Project |
First Prize | 计信未来 | North University of China | Cloud-edge Collaborative Multi-Channel Video Stitching and Intelligent Analysis System |
Second Prize | 科学养猪队 | Huazhong University of Science and Technology | SOPHGO Breeding System |
Third Prize | 乘风破浪队 | Hunan University Of Technology and Business | Classroom Teaching Behavior Analysis for Edge Intelligence |
SOPHGO has developed a comprehensive computational product matrix covering the entire spectrum of "cloud-edge-device" scenarios, providing computational products and holistic solutions for applications such as urban intelligence, smart computing centers, intelligent security, smart transportation, industrial inspection, smart terminals, among others. Compared to contemporaneous processors and graphics, the TPU delivers a performance boost of 15-30 times, particularly excelling in edge scenarios. The competition primarily assesses contestants' abilities to innovate using SOPHGO's TPU platform for applications, emphasizing machine vision, detection, and recognition in settings like robotics and drones.
Participants are tasked with leveraging the TPU to implement new applications and algorithms in various domains, including but not limited to industrial safety, general parks, smart food safety, municipal management, smart power, public safety, intelligent transportation, smart coal mines, robotics, drones, machine vision, etc. They are required to construct an entire edge computing system and realize the corresponding functionalities.
The development board based on BM1684 is the foundation. It's crucial to leverage the TPU's acceleration features, showcasing its unique advantages and achieving the optimal configuration in terms of cost and performance based on scene requirements. Various devices like screens, keyboards, mice, cameras, headphones, VR equipment, etc., can be extended, creating a comprehensive edge computing workstation. Alternatively, it can be embedded into smart cars and drones for mobile terminal edge computing. Additionally, the addition of processors like FPGA as TPU's heterogeneous accelerators can accomplish more functionalities.
SophonSDK is SOPHGO's custom deep learning SDK tailored for its proprietary processor. It covers functionalities required during the neural network inference phase, such as model optimization and efficient runtime support, offering an easy-to-use, efficient full-stack solution for deep learning application development and deployment. Participants can access relevant SDK materials for learning and use:
Please log in to the official website of the China Graduate Innovation Practice Series Contest to register and sign up before the deadline for registration, and complete the payment and submission of works.
The National Finals will be held in mid-August.