Giorgos Mountrakis, Ph.D.
Founder & Director | Intelligent Geocomputing Laboratory
Professor | Dept. of Environmental Resources Engineering | State University of NY College of Environmental Science and Forestry
Associate Editor | ISPRS Journal of Photogrammetry and Remote Sensing
Multiple Ph.D. Assistantships available for August 2025
Deep Learning in Remote sensing
The Landsat archive, with a multi-decadal global coverage is a prime candidate for deep learning classification methods due to the large data volume. We used 21 million labeled annual temporal sequences to compare deep learners with traditional methods. Results indicate substantial classification improvements of deep learning methods (DLMs) over the RF. These improvements are more pronounced on challenging to classify pixels in heterogenous areas. RF classification accuracy reaches about 70% on average, while DLMs are at 86%-95% range, depending on model architecture. See more here.
Remote sensing research highlights

Creation of pixel-based accuracy maps for RS classification

Fifteen years of published land cover classifiers

Review of global and regional land cover products

Support Vector Machines in RS problems

RS multi-step classification using intermediate results

Accurate ground detection in waveform LiDAR signals

Biomass estimation using waveform LiDAR signals

A novel multi-scale regression and classification method
Interdisciplinary research highlights

Forest attrition is higher in the Western US, in public and rural lands

High urban land consumption linked to lower African American population, higher poverty rate and lower income per capita

Hurricane Katrina flooding disproportionately affected African American neighborhoods

Urban growth prediction models: Challenges moving forward
