Object-detection

This transfer learning notebook's owner is Yuki Takahashi, with the github username yuki678. All credit goes to him. I merely modified a few scripts to make it more compatible so they can be run more smoothly. Additionally, many of the scripts used to run this notebook are copyrighted by the Tensorflow authors and licensed under the Apache License, Version 2.0 (the "License").
A copy of the license can be obtained at http://www.apache.org/licenses/LICENSE-2.0. This notebook using transfer learning for object detection with models pretrained on the Coco dataset. If you would like to use your own images then after downloading this release and cloning the necessary repositories, you will eventually clone driving-object-detection. Inside driving-object-detection/images, delete all the images of traffic lights and the xml files. Upload your own images. Then label them in the object-detection bounding box format using LabelImg and upload their respective xml files. The files are saved in PASCAL VOC format. Here is a link on how to install LabelImg https://github.com/tzutalin/labelImg#installation. There are further instructions in the README file, as well as Mr. Takahashi's own README file, requirements text file, and setup environment shell script inside Mr. Takahashi's driving-object-detection repository.