Contest Overview

The Global AI Innovation Contest focuses on cutting-edge technologies and applied innovations from a global perspective with the aim to promote academic exchanges, talent training, technological development, and cross-disciplinary application and integration. It serves as a communication platform for talented individuals and seeks to build a stimulating ecosystem for the AI industry. This Contest is open to on-campus college students, developers, and researchers of all ages and nationalities. Three AI algorithm tracks are designed for the contest, and the prize for each track is CNY 500,000 in total (pre-tax)

1st Place
CNY 150,000 Cash Prizes (Pre-Tax)
Winner Certificate
2nd & 3rd Places
CNY 80,000 Cash Prizes (Pre-Tax)
Winner Certificate
4th to 6th Places
CNY 50,000 Cash Prizes (Pre-Tax)
Winner Certificate
7th to 10th Places
CNY 10,000 Cash Prizes (Pre-Tax)
Winner Certificate
In addition, 11th to 20th places will also get a winner certificate each.

Contest Tasks

Three AI algorithm tracks are designed for the contest

Track 1: Anomaly Detection Based on Medical Imaging Report

    Radiologists read and describe medical images such as CT and MRI scans. These descriptions contain substantial medical information that is significant for the application of AI in the medical field. On this track, participating teams determine whether anomalies exist in specific areas of the body and identify the categories of the anomalies based on doctors' descriptions of CT scans. In the Preliminaries, contestants are required only to detect the existence of anomalies in each area, whereas categorization of anomalies is also required for the Semifinals. Join Now >

Track 2: PANDA-based Large-scale, Multi-object Detection and Tracking

    Built by a team from Tsinghua University, gigaPixel-level humAN centric video Dataset (PANDA) is the first dataset for large-scale, long-term, and multi-object visual analysis. Captured by gigapixel cameras, videos on PANDA cover scenes with a field-of-view of 1 km². Each scene may contain a head count in the thousands with over 100× scale variation. This track is designed to encourage research on data processing algorithms for large-scale and multi-object scenarios. Join Now >

Track 3: Breeno-based Dialog Semantic Query-pair

    Breeno is a voice assistant independently developed by OPPO for OPPO, OnePlus, realme phones, and IoT devices. It provides users with fun, considerate, and convenient dialog-based services. Intent recognition is essential to a dialog system, and dialog semantic matching is one of the mainstream algorithm solutions for intent recognition. On this track, participating teams must infer whether the given desensitized short text query-pairs are of the same semantics. Join Now >


  • Jan 15 - Apr 7, 2021
    Sign-up and Team Formation
  • Mar 2 - Apr 7, 2021
    Preliminary Phase A
  • Apr 8, 2021
    Preliminary Phase B
  • Apr 15 - May 10, 2021
  • Late May, 2021
    Online Camp Ends; Final Winners Announced


phone Contact Us