Brain computer interaction provides a direct communication channel between human brains and external devices by translating human intentions into control signals. Brain-Computer Interface (BCI) systems can be developed via a variety of brain signal acquisition techniques such as electroencephalography (EEG), electrocorticography (ECoG), magnetoencephalography (MEG), positron emission tomography (PET), functional magnetic resonance imaging (fMRI), optical imaging (i.e., functional Near InfraRed (fNIR), etc. EEG signals is one of the most widely used brain signal modality for BCIs due to their non-invasiveness, high time resolution, ease of acquisition, and cost effectiveness compared to other brain signal recording methods.

To the best of our knowledge, there are no studies that tries to passively find out whether a person is enjoying a particular genre of videos, specifically from brain signals. To do this, the very first thing that one should be able to classify is, which type of video a person is watching. With the vision of building a BCI based video recommender system, we are experimenting with several state-of-the-art algorithms for each of the submodules (pre-processing, feature extraction, feature selection and classification) of the Signal Processing module of a BCI system in order to find out what combination of algorithms best predicts what type of video a person is watching.

For more information about this project, click here.


A number plate, also known as a license plate is a metal or plastic plate used to uniquely identify vehicles. The objective of an Automatic Number Plate Recognition (ANPR) system is to locate and recognize the number plate of vehicles automatically. An ANPR system can be broken down into three steps: number plate detection, character extraction and character recognition.

There are many algorithms to detect and extract the number plate from the image of a vehicle. For example, Gabor transform, edge detection based algorithms, dynamic programming, AdaBoost, etc. However, there are very few studies for number plate of vehicles registered in Bangladesh containing Bengali (Bangla) alphanumeric characters.

In this project, we are experimenting with state-of-the-art Deep Learning techniques, one of the most popular topics of Machine Learning in the recent past, to automatically detect and recognize number plates from images of vehicles registered in Bangladesh.

For more information about this project, click here.


Hangman is a word game between two or more players were the first player thinks of a dictionary word and tells how many letters it contains, and the opponent is given limited chances to guess those letters and to form the word. The opponent then guesses a letter. If the guess is incorrect, one part the hangman body is drawn. If the guess is correct then the opponent continues to offer other letters unless s/he can, within the limited chances, find out the right word that the first player thought about. Either the opponent wins if the word is correctly guessed or, s/he is ‘hanged’ (i.e. the full body of the hangman is drawn).

As there is no word game in Bengali (Bangla) for Bangla-speaking children, we decided to build the very first Bangla word game which we named ঝুলন্ত মানব (Jhulonto Manob) which is a Bangla version of the conventional word game Hangman. However, Hangman for a new language means the game playing strategy and the rules (specifically the number of guesses/mistakes/chances that the game should allow) must change both of which are dependent on the complexity of the language. Bangla is inherently a very different language than English and this may require a different strategy to design such a game for Bangla. Moreover, this design will mostly depend on the playing strategy. In this project we analyzed different types of Bangla character frequencies to be able to devise a game playing strategy that can be utilized in developing Bangla word games in the future.

For more information about this project, click here.


Mobile applications are changing every day. Modifications of an application as simple as the keypad are very common these days since users demands are changing on a regular basis due to the fast changing mobile technology. To meet the demands of the users on mobile phone apps, we ran an extensive study to identify the area of mobile apps which needs improvement.

After conducting an affinity analysis on the data gathered from a field study which was a combination of semi-structured Interview and Contextual Inquiry, we came to the conclusion that modification of the Camera app will turn out to be very interesting.

In this study, one of the most important points stated by the participants of the study were that taking out the mobile phone from the pocket/purse, unlocking it, opening the camera application and then clicking a photo is a very time consuming process when trying to capture a moment which is changing fast. This could be a child’s cute smile or his/her adorable yawn, or a fabulous fast travelling car on the road, etc.

To solve the problem stated above, we decided to build an application that would take photos as long as the user did not stop it. The application, which we named "InsCam", when launched, starts capturing photos automatically one after the other (known as Burst Mode in photography) until the user manually stops it.

For more information about this project, click here.


Contact Me