LSTM

Project [WIP]

Authors :

• Samy BOUCHNAIF
• Duc-Vinh TRAN

I. Description:

---

Talkative Bot : a bot trying to be creative. You can find me at : https://twitter.com/TheTalkativeBot

How do I work? Post a picture in comment of one of my tweets and I will tell you what your picture evokes to me :)

(In order to avoid flooding, I will need to follow you otherwise I won't answer you!)

II. Requirements:

• python 2.7
• cuda toolkit & cudnn
• tensorflow (cpu or gpu)
• keras
• opencv
• demjson
• python-twitter library

Warning:

Files for image_captionning's model such as weights and pre-trained models can't be store in Github because of their size, I will try to find a way to store them elsewhere and make them available to download.

III. Run

• src/main.py

IV. Description:

• 01-scrapers contains notebooks about the scrapers and downloader from FTP.
• 02-train_LSTM contains notebooks for the trainings ofc differents LSTM models (in French, English, German, etc.)
• 03-image_captionning contains notebooks for the training of the Image caption model.
• 04-twitter_bot contains a notebook on how to use the Twitter API to automate actions.
• src folder contains the script files necessary to run the bot.

Illustration :