Research Interests
My PhD research focuses on adapting and evaluating Earth foundation models for reliable geospatial applications. My work investigates multimodal learning from spectral and geometric data, representation transfer across sensors and geographies, and the robustness of pre-trained models under sparse-label and real-world deployment conditions. Downstream tasks such as environmental monitoring, land-cover mapping, and 3D scene understanding serve as testbeds for studying scalable and transferable geospatial AI. My research builds on previous work in hyperspectral remote sensing, methane detection, LiDAR semantic segmentation, and digital twins. Alongside my academic research, I design automated geospatial ETL pipelines, spatial data architectures, and analytics tools for large-scale telecommunications networks. This industry experience motivates me to develop methods that are not only scientifically rigorous, but also robust, scalable, and practical for production systems.
Education
- Master of Science in Geomatics, University of Waterloo, 2022-2024
- B.E.S. in Honours Geospatial Data Science with (Co-op), Computer Science Minor, University of Waterloo, 2017-2022
Recent Work
- Hu, K. J., Chai, Y., Asgarpour, S., Boudreault, R., Li, J., & Yin, S. (2026). Methane Detection of Super-Emitters by Remote Sensing and Investigation of Wind-Driven Bias in Complex Terrain: A Multi-Instrument Analysis. Mining, 6(2), 35. DOI: 10.3390/mining6020035
- Fatholahi, S. N., Yin, S., Hu, K., He, H., Yao, K. Y., Zhang, D., Lu, D., Li, J., & Teng, J. (2026). Comparative Evaluation of Deep-Learning Models for Point Cloud Upsampling: Insights from Indoor Parking Lot Data Set. Photogrammetric Engineering & Remote Sensing, 92(6). DOI: 10.14358/PERS.25-00083R3
- Hu, K. J., Shpir, M., Clausi, D., & Xu, L. (2025). Unlocking the Potential of PRISMA Hyperspectral Imagery for Precision Agriculture Mapping. 46th Canadian Symposium on Remote Sensing (CSRS).
- Hu, K. J., Clausi, D., & Xu, L. (2025). Enhancing Seal Detection in Arctic Regions through Fine-Tuning of Faster R-CNN: A Case Study Using NOAA and DFO Datasets. 46th Canadian Symposium on Remote Sensing (CSRS).
- Hu, K. J. (2024). Semantic Modelling of an Indoor Parking Garage Using Hand-held GeoSLAM LiDAR Point Clouds, MSc Thesis, University of Waterloo.
- Hu, K., Du, J., Gong, X., Ma, L., and Li, J. (2023). A Comparative Study of Semantic Segmentation using Deep Neural Networks in a GNSS-denied Underground Parking Lot, Adv. Cartogr. GIScience Int. Cartogr. Assoc., 4, 18, https://doi.org/10.5194/ica-adv-4-18-2023.
- Hu, K. and Li, J. (2023). Using Least Cost Path Analysis to Plan a New Bypass Route on Highway 401 to Mitigate Traffic Congestion and Impacts in the City of Toronto, Ontario, Abstr. Int. Cartogr. Assoc., 6, 94, https://doi.org/10.5194/ica-abs-6-94-2023.
- Hu, K. J., Mahboubi, M., Chen, Y., & Fatholahi, S. (2022). Indirect Detection of Permafrost Degradation in Eastern Alaska Using MODIS and Landsat Data. XXVII FIG Congress 2022, 11–15 September 2022, Warsaw, Poland.