The project concern the compagny where i worked. Being the only member in the team AI, I realized a project using OpenCV/Dlib/Keras/Python.

The project consist to interpret the states of the face, such as the smile, iris tracking, the estimation of the position of the head relative to the camera, etc.

It was to create a system for managing visa appointments. The project targeted three platforms:

• Desktop with Qt framework: I used C++/Qt
• Mobile (Android + IOS) with Qt framework: I used C++/QML/JS/Qt
• Web application: I with Qt used the Laravel framework

We have some models used for machine translation:

• Google Neural Machine translation (GNMT) used in Google Translate.
• Convolutional sequence to sequence (ConvS2S) used by Facebook.

The goal of this project is to do research on a new architecture, who surpass GNMT and ConvS2S, this architecture is called Transformer, and it was created by google

I used python as language, tensor2tensor library (sub-library of Tensorflow), and Google Colaboratory for training model.

I created a tutorial to how use this library and it added to the official repos. Go to pull request.

Go to the repos

There are several automatic translation architectures, such as Google Neural Machine Translation used in google translation, Convolution Sequence to Sequence (ConvS2S) used by facebook.

In this article, I compare these architectures, with the Transformer architecture created by Google, and surpass them.

I make comparisons in terms of performance and quality, as well as the complexity of these architectures.

The article is publied also in Analytics Vidhya.

Go to the article

There are several automatic translation architectures, such as google translate and facebook. In this article, I introduce the Transformer architecture to be created by Google.

Google Brain and their collaborators have published an article introducing a new architecture, the Transformer, based only on attention mechanisms. It surpasses any other Neural machine Translation (NMT) models seen before such as Google Neural Machine Translation (GNMT) alias Google Translate.

The Transformer has been able to reach a new state of the art in translation. In addition to major improvements in the quality of translation, it also allows the realization of many other natural language processing (NLP) tasks.

Go to the article

This an article about Qt framework.

Qt (pronounced “cute”) is an open-source Frameworks for creating graphical user interfaces as well as cross-platform applications that run on various software and hardware platforms with little or no change in the underlying codebase while still being a native application with native capabilities and speed.

In the article, we will discuss several points of the Qt framework.

Go to the article

Gesture recognition is a hot topic in computer vision and pattern recognition. Since the existing methods have not presented a practical compromise between the performance and the efficiency.

The project started in a compagny that wanted to create an AI capable to predict the sign language. So i want to share my experience, because it’s an interesting aspect of our society.

The idea was to construct a neural network model to classify the signs, from the images captured by a camera of phone in real time. The user can communicate with another person using sign to text. We can add after text to speech.

The article is publied also in Becoming Human.

Go to the article

This is an application consists of a global chat room. It's created with Qt/C++ using network TCP protocol for send messages.

This is a software multi-platform audio player. You can manage your album and play your music.

I used Qt/C ++ for the GUI and the FMOD library for sound management.

This application is a web browser created using Qt technology with the C++ Language. It contains the base feature for a browser such as :

• Download files.
• History.
• Bookmark.

Visit the repos

Created in a hackathon a prototype app in mobile for 72 hours with my teammates. We had created a multi-platform application that touches the business side, I did the mobile part with Qt/QML using WebSockets.

The goal of this, it's to create an app for consultation with a psychologist.

This application was created with the Qt framework using the QML language. The application consists of image processing based on the filters such as change the contrast, brightness, add some effect as blur, etc.

Visit the repos

This is a scrabble game created with Java/Javafx. It's a online game using socket !

Visit the repos

This program compare a compressed images with distance at the level of colors, textures, and trees, by apply a formulas, in order to make an analysis on these. And apply som operations such as intersection and union between 2 images.