neat-EO

Efficient AI4EO OpenSource framework

Purposes:

• DataSet Quality Analysis
• Change Detection highlighter
• Features extraction and completion

Main Features:

• Provides several command line tools, you can combine together to build your own workflow
• Follows geospatial standards to ease interoperability and performs fast data preparation
• Build-in cutting edge Computer Vision papers implementation
• Support either RGB and multibands imagery, and allows Data Fusion
• Web-UI tools to easily display, hilight or select results (and allow to use your own templates)
• High performances
• Extensible by design

Documentation:

Tutorial:

• Learn to use neat-EO, in a couple of hours

Config file:

• neat-EO configuration file

Tools:

• neo cover Generate a tiles covering, in csv format: X,Y,Z
• neo download Downloads tiles from a Web Server (XYZ or WMS)
• neo extract Extracts GeoJSON features from OpenStreetMap .pbf
• neo rasterize Rasterize vector features (GeoJSON or PostGIS), to raster tiles
• neo subset Filter images in a slippy map dir using a csv tiles cover
• neo tile Tile a raster coverage
• neo train Trains a model on a dataset
• neo eval Evals a model on a dataset
• neo export Export a model to ONNX or Torch JIT
• neo predict Predict masks, from a dataset, with an already trained model
• neo compare Compute composite images and/or metrics to compare several slippy map dirs
• neo vectorize Extract GeoJSON features from predicted masks
• neo info Print neat-EO version informations

Presentations slides:

• @FOSDEM 2020

Installs:

With PIP:

pip3 install neat-EO

With Ubuntu 19.10, from scratch:

# neat-EO [mandatory]
sudo sh -c "apt update && apt install -y build-essential python3-pip"
pip3 install neat-EO && export PATH=$PATH:~/.local/bin

# NVIDIA GPU Drivers [mandatory for train and predict]
wget http://us.download.nvidia.com/XFree86/Linux-x86_64/435.21/NVIDIA-Linux-x86_64-435.21.run
sudo sh NVIDIA-Linux-x86_64-435.21.run -a -q --ui=none

# HTTP Server [for WebUI rendering]
sudo apt install -y apache2 && sudo ln -s ~ /var/www/html/neo

NOTAS:

• Requires: Python 3.6 or 3.7
• GPU with VRAM >= 8Go is mandatory
• To test neat-EO install, launch in a new terminal: neo info
• If needed, to remove pre-existing Nouveau driver: sudo sh -c "echo blacklist nouveau > /etc/modprobe.d/blacklist-nvidia-nouveau.conf && update-initramfs -u && reboot"

Architecture:

neat-EO use cherry-picked Open Source libs among Deep Learning, Computer Vision and GIS stacks.

GeoSpatial OpenDataSets:

• Christoph Rieke's Awesome Satellite Imagery Datasets
• Zhang Bin, Earth Observation OpenDataset blog

Bibliography:

• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• U-Net: Convolutional Networks for Biomedical Image Segmentation
• Deep Residual Learning for Image Recognition
• TernausNetV2: Fully Convolutional Network for Instance Segmentation
• The Lovász-Softmax loss: A tractable surrogate for the optimization of the IoU measure in neural networks
• Albumentations: fast and flexible image augmentations

Contributions and Services:

• Pull Requests are welcome ! Feel free to send code... Don't hesitate either to initiate a prior discussion via gitter or ticket on any implementation question. And give also a look at Makefile rules.

• If you want to collaborate through code production and maintenance on a long term basis, please get in touch, co-edition with an ad hoc governance can be considered.

• If you want a new feature, but don't want to implement it, DataPink provide core-dev services.

• Expertise, assistance and training on neat-EO are also provided by DataPink.

• And if you want to support the whole project, because it means for your own business, funding is also welcome.

Requests for funding:

We've already identified several new features and research papers able to improve again neat-EO, your funding would make a difference to implement them on a coming release:

• Increase again accuracy :

  • on low resolution imagery
  • even with few labels (aka Weakly Supervised)
  • Topology handling
  • Instance Segmentation

• Improve again performances

• Add support for :

  • Time Series Imagery
  • StreetView Imagery
  • MultiHosts scaling
  • Vectors post-treatments
  • ...

Authors:

• Olivier Courtin https://github.com/ocourtin
• Daniel J. Hofmann https://github.com/daniel-j-h

Citing:

  @Manual{,
    title = {neat-EO} Efficient AI4EO OpenSource framework},
    author = {Olivier Courtin, Daniel J. Hofmann},
    organization = {DataPink},
    year = {2020},
    url = {http://neat-EO.pink},
  }