Registered teams for both Track 1 and Track 2 will be provided access to a web-based annotation system and will collaboratively annotate a random sample of key frames from the data sets. Annotating a key frame involves drawing several polygons in a photograph and associating each polygon with one or more of a given set of classes. Each team will be responsible for annotating up to 10000 key frames. The quantity and quality of provided annotations will play an important part in choosing teams that can compete in Track 1 and Track 2. The annotation task will end on July 5.
Information provided in the registration document and an evaluation of each teamfs annotations will be used to choose up to 10 teams to compete in Track 1 and up to 10 additional teams in Track 2. The teams chosen to compete in Track 1 will get free access to GPU cloud computing facilities. Teams competing in Track 2 will get access to a cloud instance for prototyping their solutions.
All Track 1 and Track 2 teams will be provided access to the raw video footage dataset, the full set of annotations created by all teams, and access to NVIDIA Jetson TX2 edge devices for demonstrating their results at the conference. Downloading the dataset and/or the set of annotations will not be allowed. Using the given resources, competitors can then work on their respective Track submissions.
Track 1 and Track 2 Submissions and Results will be due on this day.
Track 1 and Track 2 teams will be provided NVIDIA Jetson TX2 edge devices to deploy their models (Track 1) and solutions (Track 2) on, to demonstrate their results at the conference.
Track 1 and 2 teams will present and demonstrate their solutions at the conference.
Participants in the IEEE Smart World NVIDIA AI City Challenge are also encouraged to participate in the Challenge being held at the International Workshop on Traffic and Street Surveillance for Safety and Security https://iwt4s.wordpress.com/challenge/
For any questions, please email nvidiaAICitychallenge@gmail.com.
© 2017 Designed by SJSU Conference Team