Skip to content

Building a CNN model that correctly classifies the geolocation of Google Street View imagery, inspired by the online game of GeoGuessr. This project is completed as part of APS360: Applied Fundamentals of Machine Learning at the University of Toronto.

Notifications You must be signed in to change notification settings

NOBODIDI/APS360_GeoGuessr_NN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

46 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

APS360_GeoGuessr_NN

Directory Structure

├── baseline/
│   ├── KNN_model.ipynb
├── data_processing/: data acquisition and processing code
│   ├── Creating_Train_Val_Test_Split.ipynb
│   ├── api_request_signature.py
│   ├── geocell_imagery_sampling.py
│   ├── imagery_download.py
├── primary_model/
│   ├── TRAINING_TEMPLATE.ipynb
│   ├── /hyperparameter-tuning: includes all of the model training results
├── testing/: distance computation and output visualization code
│   ├── Collect_Image_Filenames_in_Dataset.ipynb
│   ├── Distance_Computation_and_Visualizations.ipynb
│   ├── Confusion_Matrix.ipynb

How to run

  • Download the dataset or create a shortcut to Drive at 32k_train_val_test
  • Open the primary model/TEMPLATE.ipynb file in Google Colab. Click Run All. Enjoy.

Model

This ML project is based off the ResNet 50 architecture available on PyTorch.

This is an image

Video Demo of AI GeoGuessr

IMAGE ALT TEXT

About

Building a CNN model that correctly classifies the geolocation of Google Street View imagery, inspired by the online game of GeoGuessr. This project is completed as part of APS360: Applied Fundamentals of Machine Learning at the University of Toronto.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •