Curriculum Vitae

Education

• 2015-2019, Ph.D. in Signal and Image Processing, IMAGES/LTCI, Télécom ParisTech / University Paris-Saclay
• 2012–2015, Master of Science, Department of Spatial Information and Digital Technology, China University of Mining and Technology (CUMT)
• 2008–2012, Bachelor of Science, Department of Spatial Information and Digital Technology, CUMT (Ranking : 4/160)

Vocational experience

• December 2020 - Present, Machine learning researcher, Deep Planet, LONDON, UK
  • Using AttentionU-Net and high-resolution opticalremote sensing images to find Frankincense trees in the Somaliland, with a small number of training sets, the detection accuracy can reach 90%. This task was completed in cooperation with researchers from a computer vision group at Oxford University.
  • Multi-sensor time series missing value imputation with attention Conv-BiLSTM network. With the processed data, we can monitor vegetation with high-precision (10m) and high-frequency (5 days) data during the rainy season.
  • Improved the Savitzky-golay time series filtering method. The improved method allows us to retain useful temporal features.
  • Based on historical yield information, we used the Random Forest method to forecast the yield of grapes.
• July 2019 - November 2020, Research Fellow, University College London, LONDON, UK
  • Time series missing value imputation (regression, RF, LSTM/BiLSTM) with multisensor data
  • Multisensor time-series images (SAR/optical) fusion with 3D coherent radiative transfer model
  • Time-series forecasting with dual-stage attention-based RNNs
  • Algorithm evaluation over Google Cloud
• May 2015 - November 2015, Research assistant, The Chinese University of Hong Kong, HONG KONG, China
  • Time-series images processing and software development
  • Phase unwrapping with Minimum Cost Flow (MCF) optimization method

Main Skills

• Programming Python: (NumPy, SciPy, Pandas, Matplotlib, scikit-learn), C++ (Armadillo), MATLAB, JavaScript
• Machine learning frameworks: Scikit-learn, XGBoost, Keras, TensorFlow, PyTorch, MATLAB Deep Learning Toolbox
• Operating systems: GNU/Linux (Debian/ubuntu), Windows, macOS
• Cloud platform: AWS services(AWS Lambda, EC2, S3), Google Cloud (GCP/GEE/Colab)
• Data base: SQL
• Other: Docker, Git, Bitbucket, MLflow, Postman, JIRA, Postman, Confluence, LATEX, OpenOffice, ArcGIS

Publications

[1] Combining multitemporal optical and SAR data for LAI imputation with BiLSTM network, Zhao, W., Yin, F., Ma, H., Wu, Q., Gómez-Dans, J.L., Lewis, P.E., 2020. arXiv:2307.07434

[2] Ratio-based Multi-temporal SAR Images Denoising: RABASAR, Zhao, W., Deledalle, C.A., Denis, L., Maître, H., Nicolas, J-M. and Tupin, F. IEEE Transactions on Geoscience and Remote Sensing. 2019: 3552-3565.

[3] Multitemporal SAR images change detection and visualization using RABASAR and simplified GLR, Zhao, W., Denis, L., Maître, H., Nicolas, J-M. and Tupin, F., 2019. arXiv:2307.07892

[4] Patch based adaptive temporal filter and residual evaluation, Zhao, W., Riot, P., Maître, H., Nicolas, J-M. and Tupin, F. (results can be seen on the website)

[5] RABASAR: A FAST RATIO BASED MULTI-TEMPORAL SAR DESPECKLING, Zhao, W., Deledalle, C.A., Denis, L., Maître, H., Nicolas, J-M. and Tupin, F. IEEE International Geoscience and Remote Sensing Symposium (IGARSS’2018). Valencia, Spain. July 2018.

[6] Décomposition de séries temporelles d’images SAR pour la détection de changement, Lobry, S., Denis, L., Tupin, F. and Zhao, W. Traitement du Signal (GRETSI, Lavoisier)

[7] Urban area change detection based on generalized likelihood ratio test, Zhao, W., Lobry, S., Maitre, H., Nicolas, J.M. and Tupin, F. 9th International Workshop on the Analysis of Multitemporal Remote Sensing Images (MultiTemp’2017). Bruges, Belgium. June 2017.