PL:AI - Playground Library for Artificial Intelligence

Code-Snippets, tutorials and librarys for AI algorithm development

Installation

Cloud Computing

If you do not want to setup the environment on your computer, you can also execute your jupyter-notebooks on:

• Google Colab https://colab.research.google.com/drive/
• Kaggle https://www.kaggle.com/
• AWS https://aws.amazon.com/

Clone repo

git clone https://github.com/maximilianharr/plai plai

Build and Run Docker Container

cd docker 
docker-compose build
docker-compose up

Optional: If you want to use pandas_ml (see "Known Bugs")

Latest versions of pandas and scikit-learn prevent import of pandas_ml v.0.6.1. To fix this, install supported versions

python3 -m pip install pandas==0.24.0
python3 -m pip install scikit-learn==0.20.0

Unit-Tests

Run all unittests using

python3 ${PATH_TO_WS}/common/tools/run_all_unittests.py

Misc

Presentation about AI

./common/tutorials/README.md https://docs.google.com/presentation/d/1TeR-_v_WoZFvz8AkOEy-XYhMKhS0AVy31fzfF2hcAQQ/edit?usp=sharing

AI Hands-On Tutorials

• "Machine Learning von A-Z: Lerne Python & R für Data Science!" by Jannis Seemann
  https://www.udemy.com/course/machine-learning-komplett/
• "Deep Learning, Neuronale Netze & AI: Der Komplettkurs" by Jannis Seemann
  https://www.udemy.com/course/deep-learning-und-ai/
• "Complete Guide to TensorFlow for Deep Learning with Python" by Jose Portilla"
  https://www.udemy.com/course/complete-guide-to-tensorflow-for-deep-learning-with-python/

CNNs

https://pypi.org/project/darknetpy/ https://pypi.org/project/yolov3/

Further information / links

https://neurohive.io/en/
Implementations of common nets: https://github.com/keras-team/keras-applications
Trained common nets: https://keras.io/api/applications/

Popular datasets

Some popular datasets are listed at https://www.tensorflow.org/datasets/catalog/overview
Coco (Common Objects in Context) http://cocodataset.org/#home Openimages https://storage.googleapis.com/openimages/web/index.html

Jupyter-Notebooks

Start jupyter-notebook

jupyter-notebook

Convert *.ipynb to *.py

jupyter nbconvert --to=python --output-dir=./ ${SCRIPT}.ipynb

Jupyer dark mode

pip install jupyterthemes
jt -t chesterish

Frameworks

https://github.com/facebookresearch/detr

Known Bugs

pandas_ml import error

Problem:
pandas_ml v.0.6.1 supports pandas v0.22.0 and scikit-learn 0.20.0 (1) and will throw "AttributeError" if newer version of pandas and scikit-learn provided.
(1) https://pandas-ml.readthedocs.io/en/latest/
Solution (not tested):

python3 -m pip install pandas==0.24.0
python3 -m pip install scikit-learn==0.20.0

Unfortunately this does not work yet:

sudo apt install rustc
pip install --upgrade pip setuptools wheel
python3 -m pip install bottleneck
python3 -m pip install darknetpy
python3 -m pip install yolov3