VIVA Traffic Light Detecion Challenge 2015: CALL FOR PARTICIPATION

Aims and Scope:
Traffic lights are used to safely regulate the traffic flow, and are therefore a vital part of any intelligent car, whether it is fully autonomous or contains Advanced Driver Assistance Systems (ADAS). Traffic light detection has lacked a large dataset with varying conditions to determine the current state of traffic light detection. This challenge introduces a new traffic light dataset with both day- and nighttime scenarios, which will be used to benchmark participants.

The Challenge: The challenge is twofold and consist of detecting six superclasses in either day- or nightime. A superclass corresponds to a type of traffic light. The six superclasses, or types of trafic lights, are: green, yellow, red, green left, yellow left, and red left. The traffic lights are captured in La Jolla and Pacific Beach in San Diego, California, USA. A training set of more than 40,000 annotated daytime and 14,000 annotated nightime images are supplied, and the winning entry will be the one which performs the best across all six superclasses on either 21,452 daytime or 14,297 nighttime frames.

Submission: Participants are encouraged to submit detection results for all six superclasses in both day- and nighttime scenario. The winning entry for both scenarios will be the one with the highest mean AUC for PR-curves across the six superclasses.

Participation: If you are interested and planning to participate in the VIVA Traffic Light Detection Challenge, please visit the Traffic Light Detection page. If you have any further query regarding the challenge, send the organizers an email at viva.trafficlight@gmail.com .