Pinned Loading

1. Stop-and-Search-LondonStop-and-Search-LondonPublic

   Replication and extension of Suss & Oliveira (2023) Economic inequality and stop and search in London, using 2025 Metropolitan and City of London Police data with spatial regression models.

   Jupyter Notebook

2. Image-DenoisingImage-DenoisingPublic

   Deep learning image denoiser using U-Net. Achieved 33.95 dB PSNR with targeted data augmentation to address class imbalance (+6 dB improvement on worst cases). Built with PyTorch.

   Jupyter Notebook 1

3. DeepZ-Photometric-Redshift-EstimatorDeepZ-Photometric-Redshift-EstimatorPublic

   Deep learning framework for photometric redshift estimation from HSC galaxy images using a custom Inception CNN with CRPS loss and calibrated PDF output

   Python

4. Sentinel-Flood-MapperSentinel-Flood-MapperPublic

   Flood extent mapping from Sentinel-1 SAR imagery using a U-Net with pretrained EfficientNet-B0 encoder. Trained on Sen1Floods11 and STURM-Flood datasets. Python, PyTorch, segmentation-models-pytorch.

   Python