Traffic lights detection (red and green) and recognition (circular, arrow in three direction).
Image processing and machine learning are used. Performance of PCANet and HOG features are compared. Multi-object tracking and moving average technics are used for smoothing the output.
Publication:
Zhilu Chen; Xinming Huang. “Accurate and Reliable Detection of Traffic Lights Using Multi-Class Learning and Multi-Object Tracking.” in IEEE Intelligent Transportation Systems Magazine, vol. 8, no. 4, pp. 28-42, winter 2016.
Post time: Feb-23-2017