Downloading data, posting analysis results can be done from the information page of each division.
Overview
While some say that Moore’s Law, a rule of thumb in the field of IT technology that has been playing an important role in development of IT, has been stagnating, it is expected that demand for IoT devices that are able to immediately response to users’ commands will grow amid the arrival of the IoT society. In this trend, industries have been requesting the realization of innovative AI edge computing, a technology that enables handling such response taking AI technologies.
Against this backdrop, METI decided to embark on a contest program titled AI Edge Contest aiming to discover outstanding computer technologies, human resources and ideas as well as to encourage promising new human resources to enter these fields for the purpose of establishment of innovative AI edge computing technology. Under this program, entries will use real data and compete in developing solutions to challenges by focusing on future implementations.
In the first contest this year, entries will focus on a subject titled “autonomous driving and mobility service,” one of the priority fields that Japan should tackle under the “Connected Industries” policy, and will compete with each other in accuracy of detecting target objects in images for image recognition, a technology indispensable for the realization of autonomous driving technology.
METI plans to continuously conduct a series of contests under this program targeting not only algorithms and other software, but also hardware and implementation thereof, as part of its efforts for achievement of innovative AI edge computing technology.
Division
The contest will include two divisions: Object Detection and Segmentation. Contestants may only participate in one division.
Division | Object n Detection Division | Segmentation Division |
Task | Create an algorithm to detect a rectangular region including an object from a vehicle front camera image. | Create an algorithm to segment at the pixel level the region corresponding to an object from a vehicle front camera image. |
Provided data | (Train/Test) Vehicle front camera image (Train) Labeled rectangular region of an object (Train) Image meta information (route / time of day) | (Train/Test) Vehicle front camera image (Train) Split-labeled region corresponding to an object at the pixel level (Train) Image meta information (route / time of day) |
Awards* | 1st Prize: The Ministry of Economy, Trade and Industry Commerce Information Policy Bureau director award / The Extreme Edge Award / The 1st Prize Trophy / 500,000 yen / NVIDIA TITAN V 2nd Prize: The Extreme Edge Award / The 2nd Prize Trophy / 300,000 yen / NVIDIA TITAN V 3rd Prize: The Extreme Edge Award / The 3rd Prize Trophy / 100,000 yen / Google Cloud Platform Coupon (100,000 yen) Idea Award: Idea Award Trophy / 100,000 yen | 1st Prize: The Ministry of Economy, Trade and Industry Commerce Information Policy Bureau director award / The Extreme Edge Award / The 1st Prize Trophy / 500,000 yen / NVIDIA TITAN V 2nd Prize: The Extreme Edge Award / The 2nd Prize Trophy / 300,000 yen / NVIDIA TITAN V 3rd Prize: The Extreme Edge Award / The 3rd Prize Trophy / 100,000 yen / Google Cloud Platform Coupon (100,000 yen) Idea Award: Idea Award Trophy / 100,000 yen |
Objects Identified | Car, Pedestrian, Truck, Bicycle, Signal, Signs | Car, Pedestrian, Signal, Lane |
Evaluation Method | ・Quantitative evaluation based on mAP@IoU=0.75 ・Qualitative evaluation based on model document | ・Quantitative evaluation based on IoU ・Qualitative evaluation based on model document |
* As other benefits, Participation in the Japan Automotive AI Challenge, where algorithms developed by participants will be installed in self-driving go-karts and used in a test-track competition. Up to Four teams, comprising top-placing persons who wish to participate, can take part in the Challenge.Please see the Committe・Other for details.
* The idea award is only for report submitters. Also, there may be no applicable persons.