Mapping Poverty with Computer Vision
Stanford AI4ALL 2019

Prerequisites

First, create a virtual environment. python3 -m venv .env

Activate the virtual environment. source .env/bin/activate

Navigate into the notebooks/ folder. cd notebooks/

Install the dependencies. pip install -r requirements.txt

Instructions for Day 1

First, open the terminal and run the following commands:

cd ~
git clone https://github.com/CaesarPan/stanford-ai4all-vision.git #(some of you have already done this in the class)

Now if you type ls, you should see a folder called stanford-ai4all-vision (this is where we're going to put all our work in the next two weeks!)

Here for a quick wrapup, cd directory_path is how you move into another folder (you should replace directory_path with the path of the folder you wanna go); while ls is the command for listing all the file names under the current folder you're at.

After you've done all of these, you now want to navigate into our work folder! Let's do this by cd stanford-ai4all-vision.

Now you're inside of this repo, just type jupyter notebook to do the same magic we did in the class. (remember to start jupyter notebook inside stanford-ai4all-vision folder!)

You're ready for the homework at this point. Get started!

You can find the slides for Day 1 at: https://docs.google.com/presentation/d/1qdbc4dieya-2zOC5T4PHt3AZvdbTChb2GUlY6Hgyyag/edit?usp=sharing

Instructions for Day 2

Important: DON'T use any git command from today!!

First go to our shared Google Drive and enter our folder Mapping Global Poverty. The slides are under the Slides folder and will be continuously updated.

Then for Day 2's homework, first download everything under the Homework -> Day 2 folder (including one machine_learning folder and one utils folder as well as a machine_learning.ipynb file).

Next, put all of these stuffs under the folder notebooks, which is under stanford-ai4all-vision you got yesterday (note that you already have a utils folder, so just copy paste all files under the new utils you just downloaded into the original utils folder).

Now you should have everything set up. For today's homework you just need to finish machine_learning.ipynb, which requires you to finish several functions defined in machine_learning/model_helpers.py.

Have fun!

Instructions for Day 3

Today's homework is to finish image_ops.ipynb which is under Day 3 folder in our shared Google Drive.

Also you can find the numpy_tutorial.ipynb we used in today's work session in the same place.

Project Instructions for July 6 (no idea which day is this)

The main task today is to get all the data ready. Concretely, you want to load in all the data and split them into different sets. If you get these done, you can try to construct a fully-connected neural network as a baseline model.

All the data is in notebooks/data/assorted_images/satellite_images.h5. You probably want to do some search about how to play with h5 files and how to index the data in it.