Xiaosheng ZHU

B.Sc., M.Sc.
Ph.D. Student at the Hong Kong Polytechnic University and MIT SENSEable City Lab
LinkedIn · GitHub · Google Scholar · ResearchGate · ORCID
PolyU e-mail · MIT e-mail · Personal e-mail

Professional Skills & Certifications

  • Programming

    • Python
      Data analysis, data mapping, image processing and crawler programming
    • C/C++
      Algorithms, co-programming with Python, large-project programming with UI design (Qt, STL)
    • Machine Learning
      Deep learning model design, model optimization
      Microsoft Certified: AI Services Provider (Credly badge no. 2260462d-d155-49e7-80ed-efb07cdfedc4)
    • Java & Android
      Android applications development (ArcGIS, Mapbox, Mobile GIS)
    • Web
      HTML, php, Vue.js, Bootstrap
      Certified Website Developer by Google Developer Student Club (GDSC)
  • Design and Mapping

    • GIS
      Geo-database design, Spatial Analysis, GIS system programming (ArcGIS, QGIS re-development)
    • ENVI
      Remote sensing image analysis
    • Autodesk AutoCAD
      Certified Professional Level II (certificate no. CAD21770179100278)
  • Language

    • English
      TOEFL iBT® Test (113, Feb. 2019)
      GRE® General Test (328+3.0, May 2018)


  • 🇨🇳 🇭🇰 Sept. 2015 - Jun. 2019
    Shandong University of Science and Technology & The Hong Kong Polytechnic University
    山東科技大學 & 香港理工大學
    Bachelor of Science (Geographic Info. Systems)
    [Average score 84.3/100]
  • 🇭🇰 Sept. 2019 - Jan. 2021
    The Hong Kong Polytechnic University
    Master of Science (Geographic Info. Systems)
    [GPA 3.68/4.0]
  • 🇨🇳 Jul. 2020 – Sep. 2020
    Wuhan University
    Summer School Graduate (Quantitative R.S. and Data Analysis)
  • 🇭🇰 Jan. 2021 – (Jan. 2024)
    The Hong Kong Polytechnic University
    Doctor of Philosophy (ongoing, Smart City and Machine Learning Solutions)
  • 🇺🇸 Jul. 2023 - (Jan. 2024)
    Massachusetts Institute of Technology
    Doctor of Philosophy (visiting, ongoing, LiDAR and Smart City Solutions)
    At MIT SENSEable City Lab, sponsored by the Hong Kong Polytechnic University.

Research Interests & Experiences

  • Jun. 2018
    Study on the Extraction and Evolution of Impervious Surface in Da Gu River Basin
    Supervisor: Dr. GUO Bin, Associate Professor 郭斌 副教授
    Chinese National College Student Innovation and Entrepreneurship Training Program
    中國大學生創新創業計劃入選項目 (No. 201610424013)
    Based on Landsat satellite imagery, a variety of remote sensing extraction methods for impervious surfaces are compared, and the one with highest accuracy is finally selected to be applied to the extraction of impervious surfaces in the Da Gu River Basin, Qingdao, China. Then with the analysis of the result, showing the temporal and spatial evolution of impervious surfaces in the area.
  • Aug. 2018
    A National Real Estate Database Based on Web Crawler Technology
    Supervisor: Prof. LI Yunling 李雲嶺 教授
    Using Python language and crawler-related Python libraries to design a crawler program based on the architecture of a real estate website, collecting countrywide real estate data, and then use MySQL database system to establish a national real estate information database.
  • Jan. 2019
    High-resolution Remote Sensing Image Segmentation with Convolutional Neural Network: Cloud and Building Precise Detection
    Convolutional Neural Network is a great tool for automatic image processing with the characteristic of efficient, accurate and easy-to use. Based on INRIA aerial image labelling dataset and GF-5 satellite images, I performed the test of image segmentation for high spatial resolution remote sensing images with U-net deep learning model, and the test achieved relatively good results (test accuracy > 90% for building detection on high-spatial-resolution images and test accuracy > 85% for cloud detection on hyperspectral images), indicating that the application of deep-learning technology, especially the U-net model on remote sensing image processing is feasible and reliable.
  • Sep. 2020
    Spatial-temporal Analysis and Future Risk Prediction based on Independently Collated Hong Kong Covid-19 Cases Dataset
    Supervisor: Prof. SHI Wenzhong 史文中 教授
    When the new Covid-19 epidemic hit, we constructed our own database of Hong Kong epidemic cases by interfacing with the Hong Kong government, using web crawlers and spatial data collation techniques, and conducted a spatial and temporal analysis based on this database. The results were reliable and accurate and were chosen by the Hong Kong government as reference information for decision-making.
    paper · project website
  • Nov. 2021
    A High-performance and Lightweight Deep Learning Model for Real-time Autonomous Driving Data Analysis
    Supervisor: Prof. SHI Wenzhong 史文中 教授
    Real-time autonomous driving requires the use of algorithms that can analyze and give decisions in real-time, with extremely high performance, on the data returned from the large number of sensors that self-driving cars are equipped with. In this study, I designed a lightweight and efficient deep learning model structure and successfully used the trained model to beat the most effective models available in terms of accuracy and speed.


Fundings Worked For

  • National Key R&D Program of China
    Ministry of Science and Technology of the People's Republic of China
  • Otto Poon Charitable Foundation
    Work Program CD03, 1-99XK, P0035181
    Otto Poon Charitable Foundation Smart Cities Research Institute


If you think my article is useful to you, please feel free to appreciate