On behalf of QBS System Limited, we cooperate with HKIVE to supervise two groups of student from Higher Diploma in Data Science and Analytics. More details will be available in the future...
Python
On behalf of QBS System Limited, we cooperate with HKIVE to supervise two groups of student from Higher Diploma in Data Science and Analytics. More details will be available in the future...
A series of python seed/example projects, to demonstrate python developnment with different frameworks. Including Tkinter, Flask and Kivy, more frameworks will be available soon.
This project contains some util function and object which convenient the development of Computer Vision related project. Most object is developed based on OpenCV, for instance, cvplus.BoundingBox a custom object to simplify cropping and drawing operation.
An internal SDK for Computer Vision based application developnment. The SDK included three main core, Face Recognition, Object Detection and Human Pose Estimation, along with serval convenient class like VideoController to manage read and write multiple videostream via OpenCV.
An internal tool built to enable quick object detection model training with various dataset. The tool was built with Keras, detection model used YOLO-v2 as base, the feature extractor section can switch between different CNN architectures including MobileNet. The tool accept PASCAL POC format for annotation only. A supporting tagging tool was built along with the project as well.
To improve the security and safty at construction site, we cooperated with one of the Hong Kong Biggest Construction Company to build a prototype of site monitoring system. The system were able to identify and record every worker in and out at the main entrance. In the meantime, the system will also track every worker's brief location inside the construction site via floor-level tracking. Apart from regular access monitoring, the system can prevent and alert authorities whenever a working was missing as early as possible.
This project aim to simpify the procedure for passenger and ground-staff at boarding area, by using face recognition on smart glasses replace the necessary of passenger validation via passport. A finger-size barcode scanner is connected to the smart glasses with bluetooth to scan the barcode from flight ticket. Then, the user's face will be captured and sent to the server for comparison. If success, a confirmation signal will be shown on the screen. Vice versa, the correct photo registered previously will be shown to the ground staff to validate the passenger manually.
A project aim to develop a robot that performs person search in an office environment. Take advance from number of technologies, including face detection, face recognition, speech recognition, localization, path planning and re-planning in the presence of dynamic obstacles, and a decision theoretic search strategy.
The main novel contribution of the system is an ensemble method for combining the outputs of multiple classifiers for face recognition. A second contribution is the use of a strong priority over the location of individuals in the database to guide search. The system is also able to use speech generation and recognition to interact with individuals to achieve its goals, as well as to receive goals via a mobile interface.
However, source code are not available yet, due to permission from other authors.
A project aim to create a new visualisation system with a gesture interface via LeapMotion. In contrast to similar software, the project aims to deliver a system with a much user-friendly interface such that it is easy to learn.
In addition, the system maintains its own file system to eliminate duplicated work and a smooth rendering process. Nonetheless, it also allows user-customisation in several areas.
Source code and report are available on github.
WarFruit is a competitive multiplayer game, same as other popular strategy games (e.g. Command and Conquer, Age of Empire). The game was a result of the integration of the team's unique ideas along with some concepts from these other games. The game was developed via Java using Swing, LWJGL and other libraries. Besides against players, a simple AI was also introduced in the game.
Report and presentation slide are available. However, source code are not available yet, due to permission from other authors.
Outgoing is multi-platform software allows Organizer, for instance, teachers of a sightseeing group to track and communicate with its participants. Except mobile interface, a web interface was introduced to the system, allows Administrator to maintain or view varies type of datas. Or even, being notified when emergency happens.
All source code, report and presentation slide are available.