mltools 1.0.4

Pip+Conda based installation

A collection of Machine Learning (ML) Tools for object detection and classification on DG imagery.

mltools is MIT licenced.

The purpose of this repository is to enable fast prototyping of object detection and classification solutions using training data from DG Crowdsourcing (aka Tomnod).

There are four modules:

• data_extractors: functions to get pixels from georeferenced imagery;

• features: functions to derive features from pixels;

• crowdsourcing: interface with Tomnod to obtain training/test/target data and to write machine output to Tomnod DB;

• json_tools: functions to manipulate json and geojson files.

A ML algorithm (MLA) is a class with train and classify/detect functions. Presently, the repo contains the PolygonClassifier MLA which can classify a set of polygon geometries in a geojson.

An MLA is typically employed in a script which:

• retrieves training data from the Tomnod database;

• trains the MLA;

• tests the MLA and computes accuracy metrics;

• deploys the MLA for detection or classification;

• writes the MLA results back to the Tomnod database.

Example scripts can be found under /examples.

Installation/Usage

For Ubuntu, install conda with the following commands (choose default options at prompt):

wget https://repo.continuum.io/miniconda/Miniconda2-latest-Linux-x86_64.sh
bash Miniconda2-latest-Linux-x86_64.sh

For OS X, install conda with the following commands (choose default options at prompt):

wget https://repo.continuum.io/miniconda/Miniconda2-latest-MacOSX-x86_64.sh
bash Miniconda2-latest-MacOSX-x86_64.sh

Then run:

bash

so that modifications in your .bashrc take effect.

Create a conda environment:

conda create -n env python ipython numpy scipy gdal git

Activate the environment:

source activate env

Install mltools:

pip install mltools

You can now copy the scripts found in /examples in your project directory or create your own. Keep in mind that the imagery has to be in your project folder and it should have the same name as the image_name property in the geojson. Imagery in the format required by a MLA (e.g., pansharpened, multi-spectral or orthorectified) can be obtained with the gbdxtools package (https://github.com/digitalglobe/gbdxtools).

To exit your conda virtual environment:

source deactivate

Development

Activate the conda environment:

source activate env

Clone the repo:

git clone https://github.com/kostasthebarbarian/mltools

cd mltools

Install the requirements:

pip install -r requirements.txt

Please follow this python style guide: https://google.github.io/styleguide/pyguide.html. 80-90 columns is fine.

To exit your conda virtual environment:

source deactivate

Comments

mltools is developed as part of an effort to standardize MLA design and implementation.

Here is a slide with some ideas:

https://docs.google.com/drawings/d/1tKSgFMp0lLd7Abne8CdOhb1PbdJfgCz5x9XkLwDeET0/edit?usp=sharing

The vision is to employ MLA as part of a Crowd+Machine system along the lines of this document:

https://docs.google.com/document/d/1hf82I_jDNGc0NdopXxW9RkbQjLOOGkV4lU5kdM5tqlA/edit?usp=sharing