Yan Cheng

Professional Experience:

• PhD Researcher in Computer Vision for Earth Observations, University of Copenhagen, Denmark (2021 - Expected 2025): Pioneered computer vision applications in Earth observations to enhance environmental monitoring frameworks. Developed and improved a deep learning model for aerial photo analysis by 109%, significantly enhancing model performance. Collaborated with top international research institutions such as ETH Zurich and the USGS, fostering advanced cross-disciplinary research capabilities. Authored impactful publications and presented research outcomes at global conferences. Additionally, contributed as a Teaching Assistant in "Machine Learning for Earth Sciences," achieving high student evaluations.
• Remote Sensing Specialist & Consultant, DHI A/S, Denmark (2022 - 2024): Spearheaded cross-functional environmental projects, optimized multi-sensor data processing, and significantly enhanced data management efficiency through automation and advanced data management strategies.
• Data Analyst in Climate-Related Financial Risks, World Resources Institute, USA (2020 - 2021): Innovated machine learning methodologies for climate risk assessment that were integrated into investment banking frameworks; also facilitated Python programming workshops to enhance data capabilities across the organization.
• Research Assistant in Remote Sensing for Biodiversity, University of Twente, Netherlands (2019 - 2020): Led data collection and processing for biodiversity mapping projects using various advanced remote sensing techniques.
• Intern in Data Sciences for Earth Observations, Esri, USA (2019): Developed a high-efficiency Python package for processing Landsat time series data, contributing to the enhancement of data processing capabilities.

Key projects

Mapping Individual Standing Deadwood from Aerial Photos using Deep Learning:

• Developed an end-to-end framework using high-performance computing, significantly improving instance segmentation accuracy by 109%.
• The project's outcomes were highly regarded, with the framework being downloaded over 300 times and findings published in multiple high-impact journals.
• Your contributions to the field were recognized with invitations to present at prominent international conferences and featured in press releases.

Predicting Multi-hazard Physical Climate Risk Under Future Climate Scenarios:

• Innovated a machine learning-based method to assess physical climate risks, including flooding, which was crucial for future scenario planning.
• The methodology and its findings were published in the high-impact journal "Nature Communications Earth & Environment," contributing significantly to the field of climate risk assessment.

Retrieving Vegetation Phenology from High-Resolution Satellite Image Time Series:

• Conducted detailed data processing and time series analysis using PlanetScope and Sentinel-2 satellite data, focusing on mapping vegetation cycles.
• Improved data processing speeds by 600%, demonstrating significant enhancements in remote sensing capabilities.
• Published the findings in "Remote Sensing of Environment," one of the top journals in the field of remote sensing, thereby contributing to advancements in vegetation monitoring technologies.

PhD in Computer Vision for Earth Observations, University of Copenhagen, Denmark (Expected 2025)

MSc (distinct) in Natural Resources Management, University of Twente, the Netherlands (2019)

BSc (first-class) in Geographical Information Science, Wuhan University, China (2017)