Arranges profile images from a twitter search in a square t-sne using the Jonker-Volgenan algorithm. Features are generated from the penultimate predictions of the InceptionResNetV2VGG16 model.
#bitcoin
#cardi
1] Install the dependencies pip install -r requirements.txt
.
1b] For faster tSNE, run pip install git+https://github.com/DmitryUlyanov/Multicore-TSNE
2] Create a file credentials.ini
and fill out your information:
[TwitterCredentials]
consumer_key = ...
consumer_secret = ...
access_token = ...
access_token_secret = ...
3] Then to pull 144 (12x12) from the search obama
python scrape.py obama 144
python render.py obama 144