maelfabien / DataScienceApp

Deploying pre-trained NLP and computer vision models in a single application

Data Science App

In this app, I'll be deploying NLP, Computer Vision, Speech Processing and Generative models in a single application on Streamlit.

What's inside?

The idea is to leverage pre-trained models as well as algorithm built on my side. It's a good opportunity for me to become more familiar with all pre-trained models available in the open-source community.

The algorithms covered are:

• Natural Language Processing
  • Part-Of-Speech Tagging with SpaCy
  • Named Entity Recognition with SpaCy
  • Sentiment Classification with Flair
  • Question Answering with AllenNLP
• Computer Vision
  • Object Detection with Yolov3
  • Face Detection with Haar Cascade Classifier
  • Face Recognition with facerecognition
• Speech Processing
  • Voice Activity Detection
  • Speaker Identification
• Generative Models
  • A GAN to generate digits
  • A GAN to generate faces
  • Text-to-images

I integrate and deploy everything using Streamlit and Render.com.

How to use it?

The application looks like this:

It is currently hosted by Render.com and each push leads to a new build of the app. I am using the standard plan (2Gb RAM). If you would like to contribute, feel free to submit a PR.

To run it locally, clone this project and run :

pip install -r requirements

Rrun the app:

streamlit run app.py