Ram Prabhakar

Ram Prabhakar

Assistant Research Scientist at Johns Hopkins University

Remote Sensing

Machine learning techniques have been instrumental in advancing remote sensing algorithms, addressing the challenges associated with different data sources and improving the accuracy of land use and land cover (LULC) classification. One key area of focus is Few-Shot Cross-Sensor Domain Adaptation, particularly between Synthetic Aperture Radar (SAR) and multispectral data. SAR and multispectral data provide complementary information, but they often exhibit significant differences due to sensor characteristics. By employing few-shot learning approaches, ML models can effectively transfer knowledge from labeled samples in one sensor domain to another with limited labeled data. This adaptation enables improved classification accuracy and facilitates the integration of SAR and multispectral data for more comprehensive and accurate remote sensing applications.

Another significant aspect is improving SAR and optical image fusion for LULC classification by leveraging domain knowledge. Combining SAR and optical data offers a synergistic approach to overcome limitations in either data source alone. ML techniques can integrate domain knowledge, such as land cover characteristics and prior information, into the fusion process. By incorporating this domain knowledge, the fusion algorithm can effectively capture the distinctive features of different land cover classes, leading to enhanced classification accuracy. This integration of domain knowledge with ML-based image fusion techniques contributes to more reliable and robust LULC classification in remote sensing, enabling better monitoring of land use patterns, environmental changes, and urban development.

Reference

  1. K Ram Prabhakar et al., Improving SAR and Optical Image Fusion for LULC Classification with Domain Knowledge, IEEE IGARSS 2022.
  2. K Ram Prabhakar et al., Few-Shot Cross-Sensor Domain Adaptation between SAR and Multispectral Data, IEEE IGARSS 2022.
  3. K Ram Prabhakar et al., Multi-scale Attention Guided Recurrent Neural Network for Deformation Map Forecasting, SPIE RSS 2021.