Self-taught developer and computer vision researcher at UMaT. I build things for the web and do research in semantic segmentation and UAV remote sensing.
Currently working on cross-domain transferability of segmentation models (DeepLabV3+, SegFormer, UPerNet-Swin) across aerial imagery datasets.
Builder of RefConvertor a reference format conversion tool.
Contributing to MMSegmentation.
Interests: dense prediction, domain adaptation, full-stack development
