Yaseen's Portfolio

DoubleBench's OnePhoto application is a cutting-edge tool designed to ensure the authenticity and integrity of natural photos. Leveraging the JPEG Privacy & Security (PNS) international standard developed by DoubleBench, OnePhoto offers a robust solution for verifying photo authenticity and protecting image integrity.

Technologies: OpenCV, NDK, SDK, C++ and Java

Watch demo video

Download app apk

Key Features:

• DoubleBench’s ONE PHOTO authenticates natural photos.
• ONE PHOTO’s integrity checking fails whenever any tool touches the photo, including metadata changes after shooting.
• ONE PHOTO’s integrity checking also fails for the .jpg image not taken by ONE PHOTO.
• ONE PHOTO is based on the JPEG Privacy & Security international standard format developed by DoubleBench.

Publication: A Watermark-Based Scheme for Authenticating JPEG Image Integrity

Our Android-based application tackles common issues in image stitching, including stitching failures and distortion. Developed with the Android NDK (Native Development Kit) and Java, the application ensures precise and efficient stitching of panoramic images.The app uses a gyro sensor to automatically capture images based on orientation changes, which automates the stitching process and enhances user experience. This automation streamlines the creation of panoramic images, reducing manual effort. Moreover, the application integrates advanced privacy and security features with a signature based on the JPEG Privacy and Security standard. This ensures the integrity and authenticity of the captured images. The images provided illustrate the application's capability to generate high-quality panoramic photos, demonstrating its effectiveness in producing clear and accurate stitched images.

Technologies: Android Gyro, OpenCV, NDK, SDK, C++ and Java

Download APK

Key Features:

• Android-based application
• Address the issue of stitching failure
• Address the issue of distortion
• Control events with an orientation sensor for automation

Publication: Automatic Sequential Stitching of High-Resolution Panorama for Android Devices Using Precapture Feature Detection and the Orientation Sensor

A straightforward application developed in Java, designed to enhance vocabulary and grammar usage for Korean language learners. Based on the KiiP program syllabus, this app covers course content up to Level 3, providing users with a comprehensive tool for mastering Korean.

Technologies: Android Java

Download APK

Key Features:

• Android-based application
• Provides Korean Vocabulary based on KiiP Books
• Korean Grammar Usage

Initially, static hand gestures were classified using models trained on HGR datasets. For dynamic gestures, more advanced techniques such as LSTM, MediaPipe, and Transformers were employed for video-based hand gesture classification. These models achieved high accuracy, making it a robust system for real-time drone control through gesture commands.

Technologies: Python, Mediapipe, Tensorflow, Pandas, Numpy Arrays etc.

Watch demo video

Key Features:

• Used MediaPipe, TensorFlow, and Python for model development.
• The vision-based model was developed using state-of-the-art techniques to control the drone in real-time.
• The system was successfully interfaced with a drone simulation in Unity.

Publication: Next-Gen Dynamic Hand Gesture Recognition: MediaPipe, Inception-v3 and LSTM-Based Enhanced Deep Learning Model

Allowing the drone to receive and execute real-time commands via socket communication. The drone performs actions such as moving up, right, left, forward, as well as taking off, landing, and halting on destination PC while gesture were performed over a remote PC. The communcations between two PCs was established via ip addresses using sockets and TCP protocol.

Technologies: Python, Tkinter library, sockets

Watch demo video

Key Features:

• Developed a desktop based user interface
• Integrated hand gesture recogitnion model
• Sokcet communication were used for sending command over the internet using TCP
• Three modules were integrated together Unity simulator, GUI interface, machine learning HGR model

Publication: Submitted to Drones MDPI

This project showcases responsive drone control, simulating real-world conditions. Technologies used: Unity, C#

Technologies: Unity Engine, C#

Key Features:

• Desktop-based application
• Simulating drone moments
• Recieving commands via ip address for control

More details can be found on my GitHub page.

Technologies: Channel sampling software tools, MATLAB

Key Features:

• For heart disease classification

Publication: Classification of Heart Sound Signal Using Multiple Features

This dataset was created specifically for hand gesture recognition (HGR) to control drones in real time. It includes a comprehensive collection of both static and dynamic hand gesture data, designed to train and benchmark deep learning models. The dataset will be accompanied by a research paper detailing the data collection process, model training, and application in drone control. This makes it an essential resource for researchers in gesture recognition and autonomous drone systems.

Technologies: RGB camera

Key Features:

• Designed and created a benchmark dataset of hand gestures
• Will be publicly available soon

Publication: Manuscript writting under progress ...