This repository contains a Python-based machine learning project for predicting building heights. The project uses PyTorch, PyTorch Lightning, Optuna, and Google Street View API for data downloading, processing, model training, and prediction.

The project is structured as follows:

• download_massings.py: Downloads a dataset of 3D building massings from the Toronto Open Data portal, unzips the data, and processes it into a pickle file.
• download_streetview.py: Uses the Google Street View API to download images of buildings. It keeps track of which buildings have been photographed and saves the images to a specified directory.
• download_cityscapes.py: Downloads and processes the Cityscapes dataset, a popular dataset for semantic urban scene understanding tasks in computer vision.
• predict.py: Loads a trained model and a dataset, makes predictions on the dataset, and saves the predictions to a CSV file. It also reports some metrics about the predictions, such as Mean Absolute Error (MAE) and standard deviation.
• train.py: The main training script. It loads the data, applies transformations, and trains a model using PyTorch Lightning. It also supports hyperparameter optimization using Optuna.
• building_dataset.py: Defines the BuildingDataset class used in the training and prediction scripts.
• building_model.py: Defines the BuildingModelSm class used in the training and prediction scripts.

• Python 3.8+
• PyTorch
• PyTorch Lightning
• Optuna
• Google Street View API

1. Clone the repository.
2. Install the required packages.
3. Run the download_massings.py, download_streetview.py, and download_cityscapes.py scripts to download and process the data.
4. Run the train.py script to train the model.
5. Run the predict.py script to make predictions on the dataset.