Extracting Features of Streets of Singapore

Python
2018

For my Phd study, I needed to extract features from the streets of Singapore. For this purpose, I needed to first get photos of streets in some intervals such as every 10m. I took the photos from Google Street View using a Phyton script that is running in several Google Cloud instances. All these photos are named sequentially and stored in a repository.

For the extraction of the features, I used deep-learning and used Tensorflow as the tool. For deep-learning you first need to train your algorithm with a labelled dataset, so for this purpose I used Citiscapes dataset.

Features from Citiscapes

After training the algorithm, I input the photos that I have collected and at the end I got images that have the features extracted.