Google Code-in

TensorFlow (TF) is an open source machine learning library for research or production that can be used to recognize and unlimited number of things. We are expecting to use it to recognize monuments from different cities.

We are using TF with the Liquid Galaxy for project Dronecoria, on the Liquid Galaxy server (the one that runs the applications we develop beside the core system), we run TF models to detect burned areas of forest. If you're interested look for Dronecoria in the task repository.

Using different images (you will have to determine your the number of images of your dataset) from important monuments of different cities (start with 5 different monuments) write a TF model that recognizes that monument. Images to be used have to be copyright free. Write documentation about how the system works.

Added note: yes you can use Keras wrapper over TF.

Delivery:

• record a short .mp4 video of your computer screen showing up the model running.
• Send us the code and the manual by mail at lgcodein@gmail.com
• Publish it too in the selected task of your Google Code-in dashboard for mentor review and approval.