Binary image segmentation of traffic sign images (i.e. each pixel is classified as either background or foreground) which we approach from a transfer learning standpoint by taking a VGG16 model trained on ImageNet and fine-tuning it to our dataset (a collection of 1000 original images of traffic signs from the city of Zagreb which we manually labeled).
If you have Python 3.6 you can work directly on your machine within a Python virtual environment:
$ python -m venv venv
$ pip install -r requirements.txt
Alternatively you can run inside a Docker container:
$ docker run -it -p 6006:6006 -v $(pwd):/tmp -w /tmp tensorflow/tensorflow:latest-py3 bash
Train the model:
$ python src/train.py
Monitor training:
$ rm -f ./out/tensorboard/* && tensorboard --logdir ./out/tensorboard
Evaluate the model:
$ python src/test.py