Sentiment analysis and real-time age and gender determination

A general web application whose purpose is to clarify some applications of artificial intelligence in computer vision, such as determining age, gender, and feelings, in addition to identifying objects in general.

Tools:

Javascript, React, Tensorflow.js, Published using Python-Flask.

Methods:

SSD-500 -> Object Detection “COCO 80 Classes”.

MTCNN -> Face Detection.

Multi-task CNN -> gender, age, | Facial expression.

--Mai-Scanner #OCR API

Mai-Scanner is a web application that reads the contents of invoices from scanned pdf files and images, and saves them in databases, where it automatically locates fields,

In addition to reading the documents in several languages, including Arabic.

The application is currently used by a government agency in the Sultanate of Oman.

You can see a demo of this project.

Tools:

Python, Javascript.

TensorFlow, spaCy, Opencv, scikit-learn, Django.

Methods:

Text localization “Custom Dataset like ICDAR” -> #RetinaNet Object detection network

LSTM-GRU -> #OCR

NLU -> Analyze and understand fields within each document based on NLU,

In order to determine the appropriate place to save in the database.