The event will consist of a qualifying round and a final round.
The former will feature three tracks: non-directional algorithm competition, directional algorithm competition, and application algorithm collection. For each track, award winners will be selected based on the final algorithm results.
I. Non-directional Algorithm Competition
Based on the desensitized data provided, participating teams determine a topic on their own with potential social value in mind, complete data labeling and processing, develop algorithms, describe their application solution, and carry out model training offline. They will be assessed on business and technical levels such as social value, algorithm accuracy and innovation, and solution innovation. In the qualifying round, outstanding algorithms will be selected by the organizer to qualify for the final round.
II. Directional Algorithm Competition
The organizer sets up two types of algorithm requirements and provides test sets, and the contestants perform algorithm debugging offline. In the qualifying round, the review system of the event platform will automatically evaluate algorithm models based on objective criteria and single out algorithms for the final round.
III. Application Algorithm Collection
The organizer defines the fields and directions of the application algorithm collection, and the contestants provide algorithms suitable for the specified application scenario, design algorithm models offline, and submit them on the event platform. In the qualifying round, outstanding algorithms will be selected by the organizer to qualify for the final round.
This event aims to collect excellent algorithms for the building of smart cities. Participating teams can identify the pain points in such fields as "smart mobility", "smart law enforcement", "smart emergency response", and "smart park" and put forward solutions and software models.
Aimed at addressing practical problems, the solutions should utilize at least one machine learning algorithm and describe in detail the algorithm principles, R&D, data collection, and format requirements, as well as answers to other relevant problems. Complete project files (e.g., code and design description), demo video, and other references should also be submitted. The solutions should contain the following:
1. Analysis of pain points
2. Description of algorithm application scenarios
3. Algorithm ideas and selection
4. Data usage description: including description of the data used and data analysis and processing
5. Data collection and processing
6. Analysis of key issues and problems in the solution
7. Solution value analysis