Hi,
I am an engineer specializing on computer vision and machine learning applications. I have years of experience working on production-level computer vision based applications such as smart surveillance, object detection and tracking, semantic and instance segmentation, face recognition, person-reidentification, facial attribute detection, and various other applications.
I believe that the detection of the traffic light and identification of the traffic light sign can be done using a single CNN, given that the required datasets are available for training. If any annotation is needed to be done, I have the necessary tools and also the manpower to perform this as I have a team of engineers and other resources working on ML projects, including annotation.
I am sure I can complete this task with an acceptable accuracy. I would like to discuss more info via chat.
Regards,
Dilina.