2024
Intelligent Traffic Asset Management System Using YOLO and DeepSORT
The Intelligent Traffic Asset Management System leverages advanced computer vision and deep learning technologies, such as YOLO for object detection and DeepSORT for multi-object tracking, to automatically detect, track, and match road signs and traffic assets. This system enables the automated recognition of traffic signs, enhances real-time tracking, and provides accurate data for efficient traffic management. The system integrates various components, including image pre-processing, object detection, feature extraction, tracking, and sign matching, to streamline traffic asset management for urban infrastructure and intelligent transportation systems.
July 2024 – August 2024 · Traffic Asset Management, Computer Vision, YOLO, DeepSORT, Deep Learning
Intelligent Traffic Asset Management System Using YOLO and DeepSORT
The Intelligent Traffic Asset Management System leverages advanced computer vision and deep learning technologies, such as YOLO for object detection and DeepSORT for multi-object tracking, to automatically detect, track, and match road signs and traffic assets. This system enables the automated recognition of traffic signs, enhances real-time tracking, and provides accurate data for efficient traffic management. The system integrates various components, including image pre-processing, object detection, feature extraction, tracking, and sign matching, to streamline traffic asset management for urban infrastructure and intelligent transportation systems.
July 2024 – August 2024 · Traffic Asset Management, Computer Vision, YOLO, DeepSORT, Deep Learning
Research on Sentiment Analysis of Live Streaming E-commerce Real-time Comments Based on RCGA-NET
This project focuses on the intersection of AI, management science and e-commerce, applying deep learning to live - streaming e - commerce. I proposed the RCGA - Net model to extract sentiment features and classify evaluation objects in comments. I collected real - world data from TikTok live - streaming rooms, including comments and sales figures, to analyze the impact of comment features on sales.
July 2024 – March 2025 · Deep Learning, Sentiment Analysis, Live Streaming E-commerce, Real-time Comments, Elaboration Likelihood Model
Research on Sentiment Analysis of Live Streaming E-commerce Real-time Comments Based on RCGA-NET
This project focuses on the intersection of AI, management science and e-commerce, applying deep learning to live - streaming e - commerce. I proposed the RCGA - Net model to extract sentiment features and classify evaluation objects in comments. I collected real - world data from TikTok live - streaming rooms, including comments and sales figures, to analyze the impact of comment features on sales.
July 2024 – March 2025 · Deep Learning, Sentiment Analysis, Live Streaming E-commerce, Real-time Comments, Elaboration Likelihood Model
2023
Machine Learning-Based Pricing of Used Sailboats
This project aims to predict the listing prices of second-hand sailboats using machine learning methods, including Neural Networks, Random Forest, LGBM, and XGBoost. After data collection and preprocessing, model performance is evaluated through learning curve comparison, with LGBM chosen for its accuracy and good fit. The analysis focuses on regional effects on sailboat pricing, exploring variations by geographic area and model type. Additionally, a regional simulation model is developed to predict second-hand sailboat prices in Hong Kong based on economic and freight indicators. Statistical methods like paired sample T-tests are used to analyze pricing differences between monohulls and catamarans.
March 2023 – April 2023 · Machine Learning, Regression Analysis, Price Prediction, Second-Hand Sailboats
Machine Learning-Based Pricing of Used Sailboats
This project aims to predict the listing prices of second-hand sailboats using machine learning methods, including Neural Networks, Random Forest, LGBM, and XGBoost. After data collection and preprocessing, model performance is evaluated through learning curve comparison, with LGBM chosen for its accuracy and good fit. The analysis focuses on regional effects on sailboat pricing, exploring variations by geographic area and model type. Additionally, a regional simulation model is developed to predict second-hand sailboat prices in Hong Kong based on economic and freight indicators. Statistical methods like paired sample T-tests are used to analyze pricing differences between monohulls and catamarans.
March 2023 – April 2023 · Machine Learning, Regression Analysis, Price Prediction, Second-Hand Sailboats
Research on the Impact of AI Anchors’ Emotions on Consumers’ Purchase Intentions
This study employs scenario-based experiments to empirically analyze the theoretical mechanisms through which high-energy emotional expressions by AI anchors in live-streaming settings enhance consumers’ purchase intentions.
February 2023 – January 2024 · Live Streaming E-commerce, AI Anchors, Emotions
Research on the Impact of AI Anchors’ Emotions on Consumers’ Purchase Intentions
This study employs scenario-based experiments to empirically analyze the theoretical mechanisms through which high-energy emotional expressions by AI anchors in live-streaming settings enhance consumers’ purchase intentions.
February 2023 – January 2024 · Live Streaming E-commerce, AI Anchors, Emotions
2020
A Grayscale-Simulated Jacquard Fabric Design Method with Information Hiding Function
This project, conducted in collaboration with a textile engineering research team, proposed a novel design method for Jacquard fabrics that integrates both grayscale image simulation and hidden information embedding. The technique utilizes combinations of English letters and Arabic numerals to encode visual patterns that are imperceptible to the human eye but embedded within the grayscale representation of the fabric.
May 2020 – November 2020 · OpenCV, Anti-counterfeiting Technology, Jacquard Fabric Design, Python, Cpp
A Grayscale-Simulated Jacquard Fabric Design Method with Information Hiding Function
This project, conducted in collaboration with a textile engineering research team, proposed a novel design method for Jacquard fabrics that integrates both grayscale image simulation and hidden information embedding. The technique utilizes combinations of English letters and Arabic numerals to encode visual patterns that are imperceptible to the human eye but embedded within the grayscale representation of the fabric.
May 2020 – November 2020 · OpenCV, Anti-counterfeiting Technology, Jacquard Fabric Design, Python, Cpp