== README for seminars on CNNs ==


=== Installing libraries ===

  * [opencv-python](https://pypi.org/project/opencv-python/) - OpenCV python wrapper

  * [python-magic](https://pypi.org/project/python-magic/) - to access the libmagic file type identification library

  * [Keras](https://pypi.org/project/Keras/) - high-level neural networks API

  * [tensorflow](https://pypi.org/project/tensorflow/) - library for high performance numerical computation

  * [sklearn](https://pypi.org/project/scikit-learn/) - machine learning built on top of SciPy

  * `pip install opencv-python python-magic Keras tensorflow sklearn`

----


=== Get Bing API code === 

It is necessary for automatic image downloading based on keyword search. **Be careful because there's only 7-day free trial**
  * Visit [Microsoft Azure](https://azure.microsoft.com/en-us/try/cognitive-services/) page
  * Go to *Search APIs* section a pick *Bing Search APIs v7*
  * Chose *Guest* with 7-day free trial
  * You can sign in with your school Microsoft account
----

=== Download the framework ===
Download the framework {{ :courses:b4b36zui:seminars:ai-intro.zip |here }}
----

=== Build dataset ===
Run `python buildDataset.py` twice and each time enter different keyword term for an item to search for which will be put into dataset.
----


=== Train model ===
Run `python train.py` and you will be prompted to pick the dataset by name