Revision history [back]

You will still need vector map information to do traffic light information.

region_tlr_mxnet and region_tlr_ssd are meant to only do color recognition of the traffic light, and traffic light detection can only be done using feat_proj nodes.

Technically speaking, region_tlr_ssd can detect position of traffic light in the image. However, current implementation expects output from feat_proj to as a trigger to do recognition.