Projects

Avatar

Human facial landmark-based avatars represent a fascinating and rapidly evolving area in the field of digital interaction and virtual reality. These avatars are created using advanced facial recognition technologies that meticulously map an individual’s facial features, known as landmarks. These landmarks include key points on the face, such as the corners of the eyes, the tip of the nose, the contours of the lips, and the jawline. One of the most remarkable aspects of this technology is its ability to bridge the gap between the physical and digital worlds, offering a new dimension of interaction that is more natural and intuitive. As this technology continues to advance, it is poised to revolutionize the way we connect, play, and work in virtual spaces. Here’s a video to share:

Face Detection and Tracking

Face Attribute Detection and Tracking focuses on recognizing and monitoring various attributes associated with human faces in images and videos. This process involves identifying and tracking facial features, expressions, and characteristics, enabling a deeper understanding of the subjects within the visual content. It is a crucial component of computer vision and artificial intelligence systems, as it provides valuable insights into the characteristics and behaviors of individuals in images and videos, enabling numerous practical applications across different industries. Here’s a video to share:

Multitask Learning for Face, Facial Landmark and Head Pose

The rapid advancement of deep learning techniques has revolutionized the processing methods of various computer vision tasks, such as facial analysis, includ- ing face detection, facial landmark detection (FLD), and head pose estimation (HPE) methods. This project shows the integration of these tasks, particularly when addressing the complexities posed by large-angle face poses. The primary contribution is the proposal of a real-time multitask detection system capable of simultaneously performing joint detection of faces, facial landmarks, and head poses. This system builds upon the widely adopted YOLOv8 detection framework. Here’s a video to share:

Driver Drowsiness Detection based FLD

Driver drowsiness detection (DDD) is a critical area of research with significant implications for public safety, especially in transportation [1] and healthcare sectors. Realtime DDD can provide timely warnings to drivers, thereby reducing the risk of accidents. Various approaches have been developed for driver drowsiness detection, including physiological measures, behavioral measures, and visual solutions. Detecting driver drowsiness in real-time is crucial for reducing the risk of road accidents and fatalities. However, current facial landmark-based methods can be hindered by various conditions, including driving at night, eye closures, and extreme head poses. To address these challenges, we propose a facial landmark-based approach using a YOLO-based network called YOLOLandmark, which can simultaneously detect faces and their dense (68) landmarks while analyzing the state of drowsiness using extracted eye and mouth information. Here’s a video to share:

Intelligent human-machine interaction system

This system combines voice recognition as an entry point to perform various operations, including opening applications, controlling devices, and providing information. The development of this system requires in-depth expertise in voice recognition, natural language processing, internet connectivity, and human-machine interaction technologies. Additionally, it’s important to ensure the system’s scalability for easy integration of more functionalities and applications in the future. The system should also be adaptable to accommodate different user needs and allow for personalized settings and configurations. Here’s a video to share: