Enrich Sequence to Sequence Video to Text Captioning with Action Recognition

Vision and Language, Sep. 2020 - Dec. 2020, Link

• Proposed a novel model, S2VTACT, to capture action features from video skeleton graphs with spatial-temporal GCN
• Utilized a stack of 2 recurrent units(GRU) encoders and decoders to learn video captioning with injected action features
• Investigated video captioning and applications of natural language processing (NLP)

Identify Human-brand Co-location Network

Advised by Professor Natasha Zhang Foutz, May 2020 - Jan 2021

• Studied on identifying broad human and brand co-location networks to advise personalized recommendation
• Preprocessed large scale location GPS data and decorated POIs for check-in data in a distributed environment with PySpark
• Investigated the application of distributed computing and clustering algorithms

Secured QEMU License Plates Dection

Operating System, Sep. 2020 - Dec. 2020

• Designed Trusted clients(CA) and Trusted applications(TA) in QEMY hypervisor to pass encrypted images
• Detected license plates by neural network library, SOD, on TA to ensure the confidentiality of the input data

Students Management System

SQL Fundamental, Mar. 2018 - Jul. 2018

• Designed user log in, search and multiple students related interfaces by Java Swing
• Implemented students information entry, query, modification and delet, student graduation and exam record
• Deployed the system on WampServer