This repository contains the code, along with the code description for my attempt at creating 2 algorithms which are able to predict where an image was taken using Deep Learning. The project was implemented using Google Colab and the Anvil Web App client. The whole project consists of 2 deep learning models as well as a web app which uses both of them. The web App gives the user the ability to do two things:
- Upload an image and get the app's prediction regarding in which US state an image was taken (uses the "1-image-integration-model").
- Type of paste a link to a game of GeoGuessr and get the app's prediction regarding in which US state the user is in (uses the "4-image-integration-model").
There are two folders in this repository: "Application" and "GeoLocating-Models". The "GeoLocating_Models" folder contains the implementaion of two Deep Learning Models:
- "1-image-integration-model" contains the code for the model which is a able to predict in which US state a single image has been taken.
- "4-image-integration-model" contains the code for the model which is able to predict in which US state, a 360 degree image split into 4 different images has been taken.
The "Application" folder contains the code implementaion of the Anvil Web App:
- "Anvil-Integration" contains the code of the Google Colab NoteBook responsible for taking in data from the app and performing predictions.
- The code for the 3 anvil app classes:
- "Main" contains the code for the Main class of the Anvil App.
- "GeoGuessr" contains the code for the GeoGuessr class of the Anvil App.
- "ImageGuessr" contains the code for the ImageGuessr class of the Anvil App.
In order to create a copy of the App on Anvil Works, which includes its design, please use the following link: