Below are some examples of previous FYP reports that may be helpful when you write your reports.
| 2021 LIXGCH1 Smart Parking Lot with Real-time Space Monitoring |
In this project, the team developed a next-generation car parking system that leverages IoT and AI to digitalize car park operations, such as car navigation, online payment, automatic parking space allocation, and car park management to improve the car parking experience. The system includes a web portal for car park administrators to monitor real-time car park usage and visualize data collected by the system for decision making. It also includes a cross-platform mobile application to digitize the access control procedures and navigate drivers to particular parking spaces. The indoor navigation was powered by 23 Bluetooth Low Energy (BLE) beacons deployed in the HKUST LG5 car park. | Proposal Prog. Report Final Report |
| 2021 RO4 Using Machine Learning and Algorithmic Trading to Beat the U.S. Stock Market Index |
The aim of this project was to develop investment strategies that can achieve higher returns with lower risks than the S&P 500 Index by applying Machine Learning techniques to traditional investment strategies. Three Machine Learning-based investing strategies were tested: Harry Browne Permanent Portfolio with LSTM (HBLSTM), Stock Selection with Natural Language Processing (SSNLP), and Pairs Trading with PCA and Clustering (PTPC). From December 2004 to March 2021, the annual return and Maximum Drawdown (MDD) were as follows: S&P 500 Index: 9.9% 55.1% HBLSTM: 20.4% 18.0% SSNLP: 14.8% 33.4% PTPC: 3.0% 31.6% 3 combined: 10.0% 7.8% |
Final Report | 2021 MA1 Sound of Silence: SOSNET: Low Latency End-to-End Lip to Speech Synthesis on Smartphones |
This project consisted of two major contributions: a deep learning model capable of synthesising intelligible speech from silent video, and a cross-platform smartphone app to enable users to easily interact with the system in daily-life scenarios. The entire lip-to-speech pipeline was trained end to end, and it converged extremely well on a single speaker subset form the GRID dataset, and it generated comprehensible audio with average STOI, ESTOI and PESQ intelligibility scores of 0.43, 0.25 and 1.25, respectively. The cross-platform smartphone app recorded silent video and returned intelligible audio within 15 seconds, thus serving as a viable proof of concept for handheld lip-to-speech synthesis in the wild. (Distinguished project) | Proposal Prog. Report Final Report |
| 2020 LZ1 Liveness Detection in Facial Recognition for a Security Gateway |
In order to detect whether people's skin is real or not, spectroscopic cameras with infrared light can be used to analyze the liveness of their faces. The basic theory of this project was that the reflection of lights on different materials differs significantly. Both RGB and NIR (near-infrared) algorithms were applied in this project. An RGB algorithm was used for detecting flesh color, while an NIR algorithm was used for detecting human skin. With the combination of both algorithms, only real human faces are detected. Three spectroscopic cameras with different filters were used to capture NIR light, while a normal digital camera was used to capture visible light. Lighting equipment that emits different wavelengths of lights was also applied to enhance the reflection. Hyperspectral data, as well as three-channel RGB image data, was collected and processed to reach the results. The project was sponsored and supported by Yitu Technology | Proposal Prog. Report Final Report |
| 2020 LZ2 Smart Traffic Signals Using Deep Learning-based Computer Vision |
This project attempted to utilize deep learning-based computer vision to develop smart traffic signals that can adapt to the current traffic conditions and optimize green duration for each lane in an intersection, resulting in a reduction of the average waiting time of vehicles and improvements in the traffic efficiency. | Proposal Prog. Report Final Report |
| 2019 KWT2 Inflo: News Data Processing for Fake News Detection |
In order to develop an automated approach that can detect fake news, this system utilizes deep neural network (DNN) models to determine the veracity or category of fake news of an input news article, either with the article alone or with an accompanying piece of evidence. Various models were developed to accommodate different types of input and they were all combined to form an application programming interface for a social networking application for sharing news articles. | Final Report |
| 2019 LZ2 Pairs Trading with Machine Learning |
Pairs Trading is a simple trading strategy which involves finding a pair of stocks that exhibit similar historical price behavior, and then betting on the subsequent convergence of their prices in the event that they diverge. This project conducted rigorous studies on the existing pairs trading strategies proposed in academic papers. The team implemented and backtested three of the most popular pairs trading strategies (the Distance method, the Cointegration method with Rolling OLS and the Cointegration method with a Kalman Filter) to better understand their profitability, strengths and weaknesses under divergent market conditions. The core achievement of this project was a novel approach to perform pairs trading by adopting the Reinforcement Learning (RL) paradigm. (Distinguished project) | Final Report |
| 2018 RO4 Optimal Investment Strategy Using Scalable Machine Learning and Data Analytics for Small-cap Stocks |
Small capitalization stocks are characterised by higher volatility and higher potential returns. This project provided a platform for small cap stock trading. It included price prediction (based on multiple regression), portfolio allocation (based on user inputs and a convex optimization library based on Markowitz’s Mean Variance Theorem) and a web application. (Distinguished project) | Final Report |
| 2015 LZ1 Topic-Based Browsing of the Amazon Discussions Feedback Forum |
Utilizes latent Dirichlet allocation to automatically organize the posts on an Internet forum based on the topics the posts; displays the results on a website so that users can directly browse the forum by topic | Final Report |
| 2020 PAN1 Age of Robot Warlords: an Augmented Reality Mobile Strategy Game for Multiple Players |
This game is an AR mobile strategy game for multiple players called the Age of Robot Warlords. It was developed with the Unity3D game engine, C#, Blender, the Photon Unity Networking 2 API and Vuforia. The AR allows players to interact with real environments to make their strategies. | Proposal Final Report |
| 2019 NGOK2 OtherSide - An AR Mobile Game to Promote Using the HKUST Library |
This AR game was developed with and for the HKUST Library. It uses machine learning, and it has an extendable backend management system, which allows convenient administration of the game. Administrators can easily analyze players' performance, preferences and behavior. (Distinguished project) | Final Report |
| 2015 PAN4 Presentation Tools with Gesture Recognition and Augmented Reality |
Cool and dynamic presentation software using a Leap Motion controller and adding Augmented Reality (AR) into the presentations; includes an add-in for MS PowerPoint (Distinguished project) | Final Report |
| 2020 WIL1 A Blockchain Database Application for Connecting Domestic Helpers and Hong Kong Employers |
This project explored the possibility of building a platform for connecting domestic helpers and Hong Kong employers, backed by a blockchain database. The value we introduced is the theoretical resistance to anyone who would like to dishonestly tamper with the records in the database, including domestic helpers, employers themselves or agencies. | Proposal Prog. Report Final Report |
| 2018 PAN2 Leveraging Smart Contracts on Blockchains and Internet of Things to create a Decentralised Sharing Economy |
This project focused on the implementation of enabling interactions between IoT devices and smart contracts deployed on blockchains, built on Ethereum. The applications of objects and WiFi sharing were be covered. | Final Report |
| 2020 KWT2 Petriversity - Gamification in Education |
This system is a real-time learning game platform for learning in a university. Players are able to create questions and wait for their classmates to rate the questions. Rated questions appear when players initiate combat with AI enemies. They have to answer a question in order to defeat the enemy and complete their time in a dungeon to receive virtual or real-life rewards. The system also includes other gamification features, such as a leaderboard and an achievement board, to promote the learning experience. | Proposal Prog. Report Final Report |
| 2020 KWT3 Wireless VR Game with Gesture and Voice Recognition |
This VR game integrates with a wireless VR headset, IBM Watson, Microsoft Speech Platform, Leap Motion, an accelerometer and fan to free the player from the restraints of traditional devices and allow more "reality" to be put into the game. | Proposal Prog. Report Final Report |
| 2020 YIKE2 First-Person Computer Shooter Game with Eye Tracking and Unity |
FPS computer game which incorporates eye-tracking technology as an alternative of mouse game input; developed in Unity using C# with the Tobii Unity software development kit. (Distinguished Project) | Proposal Final Report |
| 2018 KWT2 Trespassers: A Haunted House VR Game App with Multi-Dimensional Interaction for Android Smartphones with Google Cardboard |
The aim of this project was to design an immersive VR mobile game which narrows the gap between the game quality and experience of using high-end and low-end headsets. This was demonstrated by a first person, exciting VR game, which includes multi-dimensional interactions on smartphones with Google Cardboard and a gamepad. The game latency was acceptable, and user feedback was very positive. | Final Report |
| 2013 RO2 Battle Fury: Multiplayer Game |
A prototype third-person action multiplayer game developed with Unreal Script, Maya, 3D Studio Max, Mudbox, World Machine, GIMP, Photoshop and the Unreal Development Kit | Final Report |
| 2011 MA3 Texas Holdem Poker AI Utilizing Machine Learning Techniques |
This game includes an artificial poker player, or pokerbot, that is based on the miximax algorithm, a generalization of minimax. This algorithm relies on a good model of an opponent’s strategy. The model utilizes supervised learning, namely naïve Bayes classifiers and neural networks. | Final Report |
| 2016 PSAN2 D-ENGINE: Deferred Rendering Game Engine |
This is an open source deferred rendering game engine that is simple and easy to use, so users can learn native game programming with good exposure to the low level system. Features include deferred rendering, skeletal animation and blending, real time collision detection, a GPU terrain and particles system, SIMD (Single Instruction Multiple Data) mathematics library, custom pool memory allocator, component based game engine architecture. | Final Report |
| 2015 PSAN2 Crowbar: 3D Level 'Map' Editor for Video Games |
A new editor intended to help streamline the development process of a 3D computer game level, maintain user focus on abstract tasks that are relevant to the design process, automate the subsequent operations required to achieve these tasks and assist users in constructing complex geometry | Final Report |
| 2021 LIX1 Customizable and Sustainable Clothing Web App to Support the Slow Fashion Movement |
In recent years, there has been a shift to more online shopping and more choices for customers. In this project, the team developed a web application that provides a way for customers to have creative freedom to design and customize garments online. It includes a 3D human figure visualization tool developed on Blender. The back end includes a reliable database for tailors in Hong Kong. Every time an order is placed, tailors have the freedom to commit to orders. The team had an opportunity to work closely with one tailor in Hong Kong to test their MVP website. | Proposal Final Report |
| 2020 LZ1 Liveness Detection in Facial Recognition for a Security Gateway |
In order to detect whether people's skin is real or not, spectroscopic cameras with infrared light can be used to analyze the liveness of their faces. The basic theory of this project was that the reflection of lights on different materials differs significantly. Both RGB and NIR (near-infrared) algorithms were applied in this project. An RGB algorithm was used for detecting flesh color, while an NIR algorithm was used for detecting human skin. With the combination of both algorithms, only real human faces are detected. Three spectroscopic cameras with different filters were used to capture NIR light, while a normal digital camera was used to capture visible light. Lighting equipment that emits different wavelengths of lights was also applied to enhance the reflection. Hyperspectral data, as well as three-channel RGB image data, was collected and processed to reach the results. The project was sponsored and supported by Yitu Technology | Proposal Prog. Report Final Report |
| 2019 LIX2 Machine Learning Model for Steel Surface Defect Detection |
This project used machine learning to automatically detect six different kinds of defects in steel pipe surfaces through computer vision. The system pipeline included image pre-processing, steel surface defect detection and defect classification. The project was done in cooperation with Lenovo Research Lab. (Distinguished project) | Final Report |
| 2018 LIX4 Organic Food e-catalogue Mobile Application |
This mobile application serves as a selling platform for a local farm in Hong Kong, called the White House. | Final Report |
| 2015 MU3 Mobile Quiz Engine with Social Features |
An easy-to-maintain, scalable and stable mobile quiz game that includes a large question bank of 1,500 questions in 6 categories, social features and an algorithm that can fetch questions from the server in a self-learning manner; can be further developed and monetized later by Moonfish Software Limited | Final Report |
| 2013 FR3 Export/Import Operations and Administration System |
A complex database management system for SriKrishna Logistics in Mumbai, India; handles customs clearing and forwarding activities of import and export firms and helps internal operations, HR, administration and sales personnel perform daily duties in a more efficient, automated and secure manner; distinguished project! | Final Report |
| 2013 RO3 Stock Data Market Analysis System |
A tool for an investment company, HWK, that uses pre-processing and pattern recognition algorithms to automate and optimize the price patter recognition process for six common stock chart patterns | Final Report |
| 2010 WIL2 New Airport Cashier System for Cathay Pacific Airways & Hong Kong Airport Services Ltd |
Real-life business application for Cathay Pacific Airways and Hong Kong Airport Services Ltd. | Final Report |
| 2017 PAN1 Attacks on Internet of Things (IoT) Devices |
Internet of things has become a more and more popular concept in recent years. It plays a significant role in academia, industry and our daily lives. Therefore, it is important to investigate the current IoT structure carefully to assure the security of IoT systems. This thesis reviews protocols and technologies used in IoT along with known attacks and intrusion detection systems (IDS). It then covers analysis of each element of IoT systems and identifies vulnerabilities which may be exploited by malicious attacker. In addition, it implements attacks toward a commonly-used communication technology of IoT, Near Field Communication (NFC). The performance of the attacks was evaluated by comparing attacks which try to achieve similar effects. | Final Report |
| 2016 LUO2 RPG Teaching Assistant Management System |
A website with a responsive UI to help the CSE Department professors and administrative staff assign TAs to courses each semester, including a matching program that takes into consideration a number factors, including course requirements and schedules, the preferences of advisors and the experience, research areas and preferences of PG students | Final Report |
| 2013 LUO2 A Web Portal for Hiking in HK |
Utilization of Google fusion tables and KML files for indicating itineraries on HK trails | Final Report |
| 2013 LUO3 Fast No-SQL Database with Hardware Acceleration |
GPU parallel processing to speed up some instructions of SciDB | Final Report |
| 2013 WIL1 Course Materials Assistant @HKUST |
A web based tool that employs page classification, web page filtering, web crawling and cloud computing to automatically download lecture notes, class assignments and important announcements | Final Report |
| 2011 WIL5 Integrity Constraints of Probabilistic Databases |
A novel functional dependency normal form to reduce redundancy | Final Report |
| 2012 RAYW1 Data Mining for Business Applications |
Using statistical methods to find people most likely to help with viral marketing | Final Report |
| 2010 WIL3 Visualizing User Preferences via Advanced Data Mining Techniques |
Data mining was applied to a movie dataset. With the help of the Bucket Pivot Algorithm, the system recommends movies for users to watch, based on users’ movie preferences. It also allows users to input the name of movies they are interested in watching and ask the system if they should spend time watching them or not. | Final Report |
| 2021 LZ2 Artificial Neural Network Builder - An Education Tool for High School Students |
In order to help high school students gain a basic understanding of how to develop artificial neural networks (ANNs), this system provides a drag-and-drop approach to ANN building. It is simpler than professional tools, like PerceptiLabs and Deep Learning Studio. It has a user-friendly GUI, tutorials, an assortment of datasets, mini games, and a dashboard. Feedback from 17 high school students who tried it was positive. | Proposal Prog. Report Final Report |
| 2020 KWT2 Petriversity - Gamification in Education |
This system is a real-time learning game platform for learning in a university. Players are able to create questions and wait for their classmates to rate the questions. Rated questions appear when players initiate combat with AI enemies. They have to answer a question in order to defeat the enemy and complete their time in a dungeon to receive virtual or real-life rewards. The system also includes other gamification features, such as a leaderboard and an achievement board, to promote the learning experience. | Proposal Prog. Report Final Report |
| 2020 MA1 How Good Is Your English Pronunciation? A Computer-aided Pronunciation Training System Using Deep Neural Networks |
This is a mobile application that teaches users basic phonetics by utilizing a neural-network that is able evaluate user speech and give feedback on the phoneme level in real-time. It can also keep track of their performance over time. | Proposal Prog. Report Final Report |
| 2020 NGOK3 An Automated English Vocabulary Learning Mobile Application |
This unique application provides different ways to tackle the issue of memorizing vocabulary. It allows users to read news articles based on their preferences, and it highlights vocab words help users memorize them. It also includes flashcards and graphics to enable easier memorization, as well as study mode to help users retaining the word knowledge for longer time. And it has a space repetition algorithm which calculates the forgetting time since a word was last reviewed. | Proposal Final Report |
| 2016 LIX2 HKUST Hour of Code - iOS Mobile App to Teach Children Basic Scratch Programming Skills and Concepts |
A free and interactive iPad App for children to learn how to code with Scratch; includes a carefully-designed curriculum to teach basic programming fundamentals with well-known fairy tales, guided tutorials for kids to go through to pick up basic coding skills, project templates for free coding to help kids unleash their imagination, a special Hour of Code tutorial for a quick taste of coding within just an hour and beautiful graphics and the attractive user interface (Distinguished project) | Final Report |
| 2021 LIX2 Intelligent Sterilisation Robot |
By making use of current technologies, many companies have already developed and deployed intelligent sterilisation robots (ISRs) to help and even replace human cleaners. In this project, the team aimed to build an ISR by combining features from different types of existing ISRs. It disinfects by spraying liquid disinfectant, and it analyses its surroundings using a light detection and ranging (LiDAR) sensor. There is also a user-friendly web application interface that allows real-time control the ISR. The project showed that such an ISR design is feasible and costs far less than a commercial ISR. | Proposal Prog. Report Final Report |
| 2012 PSAN1 Quadcopter |
A real DIY four-rotor flying machine, made from scratch! (Distinguished Project) | Final Report |
| 2011 SC4 Robotic RFID Stocktaking System for use with the αGate Portal Inventory Management System |
A sturdy mobile robot capable of performing automated or manual inventory stocktaking at convenient hours via RFID scanning (Distinguished Project) | Final Report |
| 2020 MXJ2 Positive Response Recommendation in Social Media Based on Human Need Analysis |
This project aimed to develop a mapping from text analysis of social media posts to human needs and then develop a model that takes such needs into consideration to make recommendations for positive responses on seeing new posts. To do this, historic social media posts were first gathered and analyzed to understand the underlying concerns and needs of users. Need profiles were then formulated and utilized as a semantical context to understand new textual social media posts. This made it possible to generate keywords for compliments according to relevance towards both users' concerns and the content of target posts. Next, happy stories could be recommended with similar underlying needs to enrich the content of responses. A prototype application was built for the entire recommendation pipeline, and then the psychological impact of positive responses was analyzed on post owners. | Proposal Prog. Report Final Report |
| 2020 SC1 Vulnerability Analysis of Deep Neural Networks |
This thesis carefully observes how adversarial samples fool DNN models to misclassify within MNIST, CIFAR10, and Tiny ImageNet. The designed defense mechanism could generally protect against a broad range of attacks under the experiments, and the results were slightly better than some state-of-the-art techniques, like Neural Network Invariant Checking (NIC) and Local Intrinsic Dimensionality (LID), for detecting adversarial attacks under 7 adversarial attacks and 12 commonly used DNN models. | Proposal Final Report |
| 2019 LZ1 Convolutional Neural Networks with Sparse Connections along the Depth Dimension |
Deep convolutional neural networks demand high computational and memory resources, so it is difficult for them to be deployed on systems with limited resources. Network pruning techniques are widely used to prune the weights and filters of a deep convolutional neural network to reduce their cost. In this thesis, we propose a method to prune a trained network based on the structure of the training data. Each layer of the network is rebuilt by analyzing the structure of the output of the previous layer. More specifically, we learn a tree-structured probabilistic model from the output using Chow-Liu’s algorithm, analyze strongly correlated features and prune the unimportant connections. The resulting model is more compact. | Final Report |
| 2017 PAN1 Attacks on Internet of Things (IoT) Devices |
Internet of things has become a more and more popular concept in recent years. It plays a significant role in academia, industry and our daily lives. Therefore, it is important to investigate the current IoT structure carefully to assure the security of IoT systems. This thesis reviews protocols and technologies used in IoT along with known attacks and intrusion detection systems (IDS). It then covers analysis of each element of IoT systems and identifies vulnerabilities which may be exploited by malicious attacker. In addition, it implements attacks toward a commonly-used communication technology of IoT, Near Field Communication (NFC). The performance of the attacks was evaluated by comparing attacks which try to achieve similar effects. | Final Report |
| 2021 RO4 Using Machine Learning and Algorithmic Trading to Beat the U.S. Stock Market Index |
The aim of this project was to develop investment strategies that can achieve higher returns with lower risks than the S&P 500 Index by applying Machine Learning techniques to traditional investment strategies. Three Machine Learning-based investing strategies were tested: Harry Browne Permanent Portfolio with LSTM (HBLSTM), Stock Selection with Natural Language Processing (SSNLP), and Pairs Trading with PCA and Clustering (PTPC). From December 2004 to March 2021, the annual return and Maximum Drawdown (MDD) were as follows: S&P 500 Index: 9.9% 55.1% HBLSTM: 20.4% 18.0% SSNLP: 14.8% 33.4% PTPC: 3.0% 31.6% 3 combined: 10.0% 7.8% |
Final Report |
| 2019 LZ2 Pairs Trading with Machine Learning |
Pairs Trading is a simple trading strategy which involves finding a pair of stocks that exhibit similar historical price behavior, and then betting on the subsequent convergence of their prices in the event that they diverge. This project conducted rigorous studies on the existing pairs trading strategies proposed in academic papers. The team implemented and backtested three of the most popular pairs trading strategies (the Distance method, the Cointegration method with Rolling OLS and the Cointegration method with a Kalman Filter) to better understand their profitability, strengths and weaknesses under divergent market conditions. The core achievement of this project was a novel approach to perform pairs trading by adopting the Reinforcement Learning (RL) paradigm. (Distinguished project) | Final Report |
| 2018 RO4 Optimal Investment Strategy Using Scalable Machine Learning and Data Analytics for Small-cap Stocks |
Small capitalization stocks are characterised by higher volatility and higher potential returns. This project provided a platform for small cap stock trading. It included price prediction (based on multiple regression), portfolio allocation (based on user inputs and a convex optimization library based on Markowitz’s Mean Variance Theorem) and a web application. (Distinguished project) | Final Report |
| 2017 LIX7 Bitcoin Price Prediction System |
This project aimed to tackle the problem of the price volatility of Bitcoin, by using supervised machine learning to predict future Bitcoin prices. The key features that affect Bitcoin prices were classified into Bitcoin transaction, social sentiments, and macroeconomic indices. These feature data were then preprocessed and fed to classification and regression models via various machine learning algorithms. (Distinguished project) | Final Report |
| 2021 KWT2 VR Content Management System for HKUST |
This system provides a novel course material search experience, including a VR interface with voice recognition capabilities. After HKUST instructors upload and provide tags for their course materials through a web interface, their students can put on a VR headset, log in, and search for course materials in a 3D view. Searching by keywords is fun and easy, and the process is streamlined through a filtering system. After finding their desired course materials, students can then download them and see the content through their VR headsets. | Proposal Prog. Report Final Report |
| 2020 NGOK2 Smart Personal Helper for HKUST Students |
This Android mobile application is a smart personal helper that integrates different schedulers for HKUST students and applies a deep learning technique to recommend suitable events to different users. It also provides a calendar, course deadline management function and a groupmate finding function. | Proposal Final Report |
| 2019 NGOK2 OtherSide - An AR Mobile Game to Promote Using the HKUST Library |
This AR game was developed with and for the HKUST Library. It uses machine learning, and it has an extendable backend management system, which allows convenient administration of the game. Administrators can easily analyze players' performance, preferences and behavior. (Distinguished project) | Final Report |
| 2018 LIX5 UShare: HKUST Ride-sharing Android App |
This app allows HKUST members to share rides to save time and avoid bus queues. Ride-sharing events can be either taxi sharing or private car sharing. Users can also join the events of car pools or create their own. The app includes a firebase database, a ride-sharing event algorithm and a fare estimation algorithm. | Final Report |
| 2020 MA1 How Good Is Your English Pronunciation? A Computer-aided Pronunciation Training System Using Deep Neural Networks |
This is a mobile application that teaches users basic phonetics by utilizing a neural-network that is able evaluate user speech and give feedback on the phoneme level in real-time. It can also keep track of their performance over time. | Proposal Prog. Report Final Report |
| 2020 MXJ1 Meaningful Metaphor Mining: Identification and Explanation |
In this project, the team conducted plentiful research on a specific domain: metaphor understanding. To understand the meaningfulness of metaphors, they collected 6,104 human-labeled meaningful pairs from large corpora, and 1295 among them were labeled as meaningful. After doing feature analysis, they collected 765 dimensions of features in three different feature categories: psycholinguistic, conceptual, and SynSet features. Then, they adopted LightGBM to complete this metaphor meaningfulness identification work as a binary classification task. After some human-in-the-loop validation, they derived an identification model with AUC up to .895. Next, they implemented three unsupervised methods to explain a metaphor word pair: a heuristic model, a BERT masked model and a fusion model. Then, they conducted a final user study to systematically evaluate these explanation models. | Final Report |
| 2020 MXJ1 Meaningful Metaphor Mining: Identification and Explanation |
In this project, the team conducted plentiful research on a specific domain: metaphor understanding. To understand the meaningfulness of metaphors, they collected 6,104 human-labeled meaningful pairs from large corpora, and 1295 among them were labeled as meaningful. After doing feature analysis, they collected 765 dimensions of features in three different feature categories: psycholinguistic, conceptual, and SynSet features. Then, they adopted LightGBM to complete this metaphor meaningfulness identification work as a binary classification task. After some human-in-the-loop validation, they derived an identification model with AUC up to .895. Next, they implemented three unsupervised methods to explain a metaphor word pair: a heuristic model, a BERT masked model and a fusion model. Then, they conducted a final user study to systematically evaluate these explanation models. | Final Report |
| 2020 MXJ2 Positive Response Recommendation in Social Media Based on Human Need Analysis |
This project aimed to develop a mapping from text analysis of social media posts to human needs and then develop a model that takes such needs into consideration to make recommendations for positive responses on seeing new posts. To do this, historic social media posts were first gathered and analyzed to understand the underlying concerns and needs of users. Need profiles were then formulated and utilized as a semantical context to understand new textual social media posts. This made it possible to generate keywords for compliments according to relevance towards both users' concerns and the content of target posts. Next, happy stories could be recommended with similar underlying needs to enrich the content of responses. A prototype application was built for the entire recommendation pipeline, and then the psychological impact of positive responses was analyzed on post owners. | Proposal Prog. Report Final Report |
| 2016 PAN3 Real-time Emotion Sensing with Google Glass |
An emotion sensing system with Google Glass, which allows users to infer the emotions of others by analyzing their facial expressions in an automated yet unobtrusive way. The inferring process is based on machine learning and support vector machine (SVM) emotion classifiers that represent six plus one distinct emotions according to Ekman’s basic emotion model. These classifiers were painstakingly devised by first manually selecting 1,806 representative images of Asian faces from 5553 images in eight different open source facial expression datasets. Then, through OpenCV’s face detector and dlib’s face landmark extractor, 18 key landmark points were identified on each face and 17 displacement vector features were calculated for each of the seven emotions. These features are fed into LIBSVM’s machine learning library for training and emotion prediction. The combined system enables real-time emotion sensing on the Google Glass with good accuracy. The levels of sensed emotions are overlaid with augmented reality (AR) on the Android mobile phone user interface as well as audibly conveyed via the tiny headset speaker in the Google Glass platform. | Final Report |
| 2021 LIXGCH1 Smart Parking Lot with Real-time Space Monitoring |
In this project, the team developed a next-generation car parking system that leverages IoT and AI to digitalize car park operations, such as car navigation, online payment, automatic parking space allocation, and car park management to improve the car parking experience. The system includes a web portal for car park administrators to monitor real-time car park usage and visualize data collected by the system for decision making. It also includes a cross-platform mobile application to digitize the access control procedures and navigate drivers to particular parking spaces. The indoor navigation was powered by 23 Bluetooth Low Energy (BLE) beacons deployed in the HKUST LG5 car park. | Proposal Prog. Report Final Report |
| 2018 PAN2 Leveraging Smart Contracts on Blockchains and Internet of Things to create a Decentralised Sharing Economy |
This project focused on the implementation of enabling interactions between IoT devices and smart contracts deployed on blockchains, built on Ethereum. The applications of objects and WiFi sharing were be covered. | Final Report |
| 2017 PAN1 Attacks on Internet of Things (IoT) Devices |
Internet of things has become a more and more popular concept in recent years. It plays a significant role in academia, industry and our daily lives. Therefore, it is important to investigate the current IoT structure carefully to assure the security of IoT systems. This thesis reviews protocols and technologies used in IoT along with known attacks and intrusion detection systems (IDS). It then covers analysis of each element of IoT systems and identifies vulnerabilities which may be exploited by malicious attacker. In addition, it implements attacks toward a commonly-used communication technology of IoT, Near Field Communication (NFC). The performance of the attacks was evaluated by comparing attacks which try to achieve similar effects. | Final Report |
| 2021 LIXGCH1 Smart Parking Lot with Real-time Space Monitoring |
In this project, the team developed a next-generation car parking system that leverages IoT and AI to digitalize car park operations, such as car navigation, online payment, automatic parking space allocation, and car park management to improve the car parking experience. The system includes a web portal for car park administrators to monitor real-time car park usage and visualize data collected by the system for decision making. It also includes a cross-platform mobile application to digitize the access control procedures and navigate drivers to particular parking spaces. The indoor navigation was powered by 23 Bluetooth Low Energy (BLE) beacons deployed in the HKUST LG5 car park. | Proposal Prog. Report Final Report |
| 2021 LIXGCH1 Smart Parking Lot with Real-time Space Monitoring |
In this project, the team developed a next-generation car parking system that leverages IoT and AI to digitalize car park operations, such as car navigation, online payment, automatic parking space allocation, and car park management to improve the car parking experience. The system includes a web portal for car park administrators to monitor real-time car park usage and visualize data collected by the system for decision making. It also includes a cross-platform mobile application to digitize the access control procedures and navigate drivers to particular parking spaces. The indoor navigation was powered by 23 Bluetooth Low Energy (BLE) beacons deployed in the HKUST LG5 car park. | Proposal Prog. Report Final Report |
| 2020 NGOK2 Smart Personal Helper for HKUST Students |
This Android mobile application is a smart personal helper that integrates different schedulers for HKUST students and applies a deep learning technique to recommend suitable events to different users. It also provides a calendar, course deadline management function and a groupmate finding function. | Proposal Final Report |
| 2020 NGOK3 An Automated English Vocabulary Learning Mobile Application |
This unique application provides different ways to tackle the issue of memorizing vocabulary. It allows users to read news articles based on their preferences, and it highlights vocab words help users memorize them. It also includes flashcards and graphics to enable easier memorization, as well as study mode to help users retaining the word knowledge for longer time. And it has a space repetition algorithm which calculates the forgetting time since a word was last reviewed. | Proposal Final Report |
| 2018 LIX4 Organic Food e-catalogue Mobile Application |
This mobile application serves as a selling platform for a local farm in Hong Kong, called the White House. | Final Report |
| 2018 LIX5 UShare: HKUST Ride-sharing Android App |
This app allows HKUST members to share rides to save time and avoid bus queues. Ride-sharing events can be either taxi sharing or private car sharing. Users can also join the events of car pools or create their own. The app includes a firebase database, a ride-sharing event algorithm and a fare estimation algorithm. | Final Report |
| 2018 PSAN1 Early Detection of Diabetic Retinopathy (DR) via a Smart Phone App for an NGO Partner in Indonesia |
In this project, the team developed a mobile app to make the DR detection algorithm developed by the previous year's FYP team run on an Android smartphone instead of a remote server with MATLAB, thus making the system more accessible in rural areas, since no Internet connection is needed. See 2017 PSAN1 below. | Final Report |
| 2017 PSAN1 Portable Application for Early Detection of Diabetic Retinopathy (DR) |
DR is an ocular manifestation of diabetes, which is very problematic for working age people in Indonesia, the 4th most populous country in the world. Although DR can usually be prevented with early detection and proper treatment, it traditionally requires a bulky and expensive screening camera and an eye specialist. Sadly, these are seldom available in many remote areas, and it's difficult to bring them to those areas. Thus, in this project, a system was developed to enable early screening. It includes a mobile app which sends retinal images captured by a portable Horus fundus camera via the Internet to a server running an algorithm that uses MATLAB for DR detection. The system can automatically detect the disease in the absence of an eye specialist, so people in rural areas can then be sent to cities for further treatment. This project was done in cooperation with Student Innovation for Global Health Technology (SIGHT) and Gadjah Mada University in Indonesia, which helped with field testing. | Final Report |
| 2016 LIX2 HKUST Hour of Code - iOS Mobile App to Teach Children Basic Scratch Programming Skills and Concepts |
A free and interactive iPad App for children to learn how to code with Scratch; includes a carefully-designed curriculum to teach basic programming fundamentals with well-known fairy tales, guided tutorials for kids to go through to pick up basic coding skills, project templates for free coding to help kids unleash their imagination, a special Hour of Code tutorial for a quick taste of coding within just an hour and beautiful graphics and the attractive user interface (Distinguished project) | Final Report |
| 2014 LUO3 An iPhone App for Hong Kong Hikers |
This app provides hikers with an all-rounded solution and a one-stop service, from trail information to event planning, including a personalized workout manager and various assistive tools. Using MySQL database, Google fusion table and push notification service, it integrates user contributed plants and blind spot recognition services, advanced trail searching, event organizations and a social network. It also includes a shortest path advisor using Dijkstra's algorithm and emergency call service. | Final Report |
| 2014 QIAN1 Price Sharing Android App for Hong Kong Supermarkets |
This app allows users to search for product information about supermarket products by scanning a barcode or using a keyword search. The product database primarily relies upon crowdsourcing to collect product prices and give ratings. (Distinguished project) | Final Report |
| 2019 NGOK2 An AR Mobile Game to Promote Using the HKUST Library |
This AR game was developed with and for the HKUST Library. It uses machine learning, and it has an extendable backend management system, which allows convenient administration of the game. Administrators can easily analyze players' performance, preferences and behavior. (Distinguished project) | Final Report |
| 2015 MU3 Mobile Quiz Engine with Social Features |
An easy-to-maintain, scalable and stable mobile quiz game that includes a large question bank of 1,500 questions in 6 categories, social features and an algorithm that can fetch questions from the server in a self-learning manner; can be further developed and monetized later by Moonfish Software Limited | Final Report |
| 2012 LUO1 iCDD - An iPhone-based Choh Dai Di Card Game |
iPhone Choi Dai Di game with the popular initial card exchange; includes beginning, intermediate and advanced AI levels of difficulty; good introduction for Cho Dai Di beginners; developed with Xcode with the cocos2d plug-in | Final Report |
| 2020 KWT3 Wireless VR Game with Gesture and Voice Recognition |
This VR game integrates with a wireless VR headset, IBM Watson, Microsoft Speech Platform, Leap Motion, an accelerometer and fan to free the player from the restraints of traditional devices and allow more "reality" to be put into the game. | Proposal Prog. Report Final Report |
| 2016 MA2 Using the Leap Motion Controller to Translate Sign Language to Speech |
This system returns speech and text when the user performs sign language in front of a Leap Motion Controller (LMC). The LMC is used to capture hand gesture images and convert them into the positional and direction information. This data is compared with the data inside the database to determine the most similar sign using a tailor-made Dynamic Time Warping (DTW) algorithm. Once a gesture has been recognized, its corresponding meaning will be presented in both audio and text format. (Distinguished project; 2016 HKUST President's Cup Gold Award!) | Final Report |
| 2015 MA1 Leap Sense: Turn Any Computer Screen into a Gesture-Assisted Touch Screen Using a Leap Motion Controller |
Utilizes a Leap Motion Controller to detect hand movement and enable gesture control and virtual touch screen functionality (Distinguished project; 2015 HKUST President's Cup Winner!) | Final Report |
| 2015 PAN4 Presentation Tools with Gesture Recognition and Augmented Reality |
Cool and dynamic presentation software using a Leap Motion controller and adding Augmented Reality (AR) into the presentations; includes an add-in for MS PowerPoint (Distinguished project) | Final Report |
| 2015 PSAN1 AirTennis: A Web-Based, Mobile Motion Controlled Console Game |
A cool game that allows two players to play virtual tennis via computer displays and "tennis raquets" that are actually smart phones with accelerometers and gyroscopes (Distinguished project) |
Final Report |
| 2018 MA3 Project Zamplify - A Sound Recognition System for Context Awareness |
Zamplify, a real-time natural sound recognition system, includes an Android app and its complementary IoT device. Together, they continuously recognize context from sound in the surroundings, and provide a customizable trigger-action mechanism that performs actions when certain contexts are detected. The core technologies used in Zamplify include a convolutional neural network for extracting sound features and a recurrent neural network (with long short-term memory cells) for modelling sequential information of audio. (Distinguished project) | Final Report |
| 2017 PAN1 Attacks on Internet of Things (IoT) Devices |
Internet of things has become a more and more popular concept in recent years. It plays a significant role in academia, industry and our daily lives. Therefore, it is important to investigate the current IoT structure carefully to assure the security of IoT systems. This thesis reviews protocols and technologies used in IoT along with known attacks and intrusion detection systems (IDS). It then covers analysis of each element of IoT systems and identifies vulnerabilities which may be exploited by malicious attacker. In addition, it implements attacks toward a commonly-used communication technology of IoT, Near Field Communication (NFC). The performance of the attacks was evaluated by comparing attacks which try to achieve similar effects. | Final Report |
| 2011 SC4 Robotic RFID Stocktaking System for use with the αGate Portal Inventory Management System |
A sturdy mobile robot capable of performing automated or manual inventory stocktaking at convenient hours via RFID scanning (Distinguished Project) | Final Report |
| 2016 SC4 Electronic Health Record System for Developing Countries |
This system facilitates normal patient record-keeping work for NGOs and volunteer medical personnel who periodically visit rural areas with no Internet conectivity. Thanks to a Rasberry Pi single board computer, a LAN can be quickly set up for doctors and medical staff to immediately retrieve, modify and add medical records of patients in rural areas. The system is a lot more compact, portable, reliable, easy to understand and easy to use than previous solutions. The Android and iOS mobile apps eliminate the need of medical teams to carry notebook computers and power supplies to remote areas. The system also includes a dedicated API for better security and future scaling up. The team worked with SIGHT from HKUST and One2One CAMBODIA, and one team member actually spent several weeks in Cambodia gathering user requirements and field testing the system. Much documentation was also provided for the benefit of real-life end users. (Distinguished project) | Final Report |
| 2015 YIKE2 Flooding Simulation with Improved Tools and Features |
Flood simulation for a real terrain over a period of time under rainfall and other causes of flooding; utilizes the Shallow Water Equation to predict the water flow and return the water heights at respective locations over a certain time frame | Final Report |
| 2010 YIKE1 Flood Prediction Using Optimized Flood Accumulation Algorithms |
Good use of OpenGL Utility Toolkit (GLUT); various terrains and flooding parameter | Final Report |
| 2021 KWT2 VR Content Management System for HKUST |
This system provides a novel course material search experience, including a VR interface with voice recognition capabilities. After HKUST instructors upload and provide tags for their course materials through a web interface, their students can put on a VR headset, log in, and search for course materials in a 3D view. Searching by keywords is fun and easy, and the process is streamlined through a filtering system. After finding their desired course materials, students can then download them and see the content through their VR headsets. | Proposal Prog. Report Final Report |
| 2018 KWT1 Developing a Virtual Reality Search Interface for Web Search |
This Virtual Reality (VR) search interface can provide a new searching experience through 3D data representation as well as innovative interactions with users. The entire system consists of a VR mobile application, a server, and a MySQL database. A web crawling algorithm implemented on the server is used to capture useful screenshots from results from the Google Search Engine. Different kinds of APIs were also used in order to facilitate the performance of the system, including the Google Custom Search, IBM Watson voice recognition and Youtube APIs. The system was tested in multiple ways so as to ensure that it could withstand large amounts of search queries within a satisfactory response time and function on various popular smartphones. | Final Report |
| 2020 KWT3 Interactive Wireless VR Game with Gesture and Voice Recognition |
This VR game integrates with a wireless VR headset, IBM Watson, Microsoft Speech Platform, Leap Motion, an accelerometer and fan to free the player from the restraints of traditional devices and allow more "reality" to be put into the game. | Proposal Prog. Report Final Report |
| 2018 KWT1 Developing a Virtual Reality Search Interface for Web Search |
This Virtual Reality (VR) search interface can provide a new searching experience through 3D data representation as well as innovative interactions with users. The entire system consists of a VR mobile application, a server, and a MySQL database. A web crawling algorithm implemented on the server is used to capture useful screenshots from results from the Google Search Engine. Different kinds of APIs were also used in order to facilitate the performance of the system, including the Google Custom Search, IBM Watson voice recognition and Youtube APIs. The system was tested in multiple ways so as to ensure that it could withstand large amounts of search queries within a satisfactory response time and function on various popular smartphones. | Final Report |
| 2018 KWT2 Trespassers: A Haunted House VR Game App with Multi-Dimensional Interaction for Android Smartphones with Google Cardboard |
The aim of this project was to design an immersive VR mobile game which narrows the gap between the game quality and experience of using high-end and low-end headsets. This was demonstrated by a first person, exciting VR game, which includes multi-dimensional interactions on smartphones with Google Cardboard and a gamepad. The game latency was acceptable, and user feedback was very positive. | Final Report |
| 2021 LZ3 A Style Transfer System for Artworks |
The combination of art and technology enables anyone to create stylized artworks and revive historically significant genres with novel contents. As an application of machine learning facing general users, this system provides a functional artistic style transfer website, generating stylized pictures based on user inputs. This website incorporates some of the latest research in neural style transfer and performs various tasks. From the "create1" page, a photo can be transferred using Cycle Generative Adversarial Network (CycleGAN) models into ten different and impressive predefined styles, each of which is elegantly displayed in the "gallery" page. From the "create2" page, a photo can be transferred to styles extracted from user-defined style images by passing through the model based on a Convolutional Neural Network (CNN), where the extent of transformation and preservation of the original colors is subject to user indication. Transferring a video into customized styles is supported on the "video transfer" page. According to a user survey, a majority of participants gave positive feedback on the design of the website and performance of style transfer functions. | Proposal Final Report |
| 2021 LIX1 Customizable and Sustainable Clothing Web App to Support the Slow Fashion Movement |
In recent years, there has been a shift to more online shopping and more choices for customers. In this project, the team developed a web application that provides a way for customers to have creative freedom to design and customize garments online. It includes a 3D human figure visualization tool developed on Blender. The back end includes a reliable database for tailors in Hong Kong. Every time an order is placed, tailors have the freedom to commit to orders. The team had an opportunity to work closely with one tailor in Hong Kong to test their MVP website. | Proposal Final Report |
| 2021 MA1 Sound of Silence: SOSNET: Low Latency End-to-End Lip to Speech Synthesis on Smartphones |
This project consisted of two major contributions: a deep learning model capable of synthesising intelligible speech from silent video, and a cross-platform smartphone app to enable users to easily interact with the system in daily-life scenarios. The entire lip-to-speech pipeline was trained end to end, and it converged extremely well on a single speaker subset form the GRID dataset, and it generated comprehensible audio with average STOI, ESTOI and PESQ intelligibility scores of 0.43, 0.25 and 1.25, respectively. The cross-platform smartphone app recorded silent video and returned intelligible audio within 15 seconds, thus serving as a viable proof of concept for handheld lip-to-speech synthesis in the wild. (Distinguished project) | Proposal Prog. Report Final Report |
| 2020 LZ1 Liveness Detection in Facial Recognition for a Security Gateway |
In order to detect whether people's skin is real or not, spectroscopic cameras with infrared light can be used to analyze the liveness of their faces. The basic theory of this project was that the reflection of lights on different materials differs significantly. Both RGB and NIR (near-infrared) algorithms were applied in this project. An RGB algorithm was used for detecting flesh color, while an NIR algorithm was used for detecting human skin. With the combination of both algorithms, only real human faces are detected. Three spectroscopic cameras with different filters were used to capture NIR light, while a normal digital camera was used to capture visible light. Lighting equipment that emits different wavelengths of lights was also applied to enhance the reflection. Hyperspectral data, as well as three-channel RGB image data, was collected and processed to reach the results. The project was sponsored and supported by Yitu Technology. | Proposal Prog. Report Final Report |
| 2020 LZ2 Smart Traffic Signals Using Deep Learning-based Computer Vision |
This project attempted to utilize deep learning-based computer vision to develop smart traffic signals that can adapt to the current traffic conditions and optimize green duration for each lane in an intersection, resulting in a reduction of the average waiting time of vehicles and improvements in the traffic efficiency. | Proposal Prog. Report Final Report |
| 2019 LIX2 Machine Learning Model for Steel Surface Defect Detection |
This project used machine learning to automatically detect six different kinds of defects in steel pipe surfaces through computer vision. The system pipeline included image pre-processing, steel surface defect detection and defect classification. The project was done in cooperation with Lenovo Research Lab. (Distinguished project) | Final Report |
| 2019 MXJ3 Automatic 3D Head Modeling from One Image |
This prototype system achieved automatic 3D human head generation from only one 2D image. The project was divided into three parts: geometry reconstruction, texture generation and hair generation. The system achieved a reasonable reconstruction time, a high level of robustness and a high degree of flexibility. It was mainly written in Python with the support of a few packages and the Blender API, and it is compatible with popular operating systems. | Final Report |
| 2018 PSAN1 Early Detection of Diabetic Retinopathy (DR) via a Smart Phone App for an NGO Partner in Indonesia |
In this project, the team developed a mobile app to make the DR detection algorithm developed by the previous year's FYP team run on an Android smartphone instead of a remote server with MATLAB, thus making the system more accessible in rural areas, since no Internet connection is needed. See 2017 PSAN1 below. | Final Report |
| 2017 MA3 English Text to Sign Language: ASL Synthesis |
This system allows normal people to communicate to deaf people by automatically converting a given string of English text into a played animation of American Sign Language, making use of various modern technologies, including natural language processing with NLTK (to parse the English Text), Kinect and Leap Motion Controller (to create the animations database), and Unity3D (to display animations for end-users). (Distinguished Project) | Proposal Final Report |
| 2017 PSAN1 Portable Application for Early Detection of Diabetic Retinopathy (DR) |
DR is an ocular manifestation of diabetes, which is very problematic for working age people in Indonesia, the 4th most populous country in the world. Although DR can usually be prevented with early detection and proper treatment, it traditionally requires a bulky and expensive screening camera and an eye specialist. Sadly, these are seldom available in many remote areas, and it's difficult to bring them to those areas. Thus, in this project, a system was developed to enable early screening. It includes a mobile app which sends retinal images captured by a portable Horus fundus camera via the Internet to a server running an algorithm that uses MATLAB for DR detection. The system can automatically detect the disease in the absence of an eye specialist, so people in rural areas can then be sent to cities for further treatment. This project was done in cooperation with Student Innovation for Global Health Technology (SIGHT) and Gadjah Mada University in Indonesia, which helped with field testing. | Final Report |
| 2016 PSAN2 D-ENGINE: Deferred Rendering Game Engine |
This is an open source deferred rendering game engine that is simple and easy to use, so users can learn native game programming with good exposure to the low level system. Features include deferred rendering, skeletal animation and blending, real time collision detection, a GPU terrain and particles system, SIMD (Single Instruction Multiple Data) mathematics library, custom pool memory allocator, component based game engine architecture. | Final Report |
| 2015 MU1 Indoor Map System |
This system allows a user to make his or her own 3D indoor map with few steps. The system will generate a customizable 3D map after the user inputs all the necessary data. All 3D models in the map can be customized by the user. The system is based on HTML5, CSS and PHP, and it follows the model-view-controller software design pattern. The database is an online database constructed in MySQL with the aid of phpMyAdmin. | Final Report |
Copyright HKUST CSE Dept. 2024
Blog template built for Bootstrap
by @mdo.
Back to top