Hello. This is a project that using deep learning to identify traffic light signals. It is built using fast.ai. A resnet with 34 layers was trained with images from Bing's Image Search API. Refer to Traffic_Light_Detector.py for the code used in the Jupyter Notebook to actually train the model. To actually run the Detector, refer to Traffic_Detector.ipynb. Run that Notebook, and make sure you have the .pkl file downloaded. By running the second box, you may upload your own images and have them classified.