// project
03 / 06
FitScanCV
Local-first body measurement prediction pipeline using paired silhouette masks and metadata.
FitScanCV predicts body measurements from paired front and side silhouette masks using the BodyM dataset. The project includes raw dataset inspection, a verified photo-level manifest, subject-safe train/validation splits, manifest-backed dataloaders, baseline CNN models, pretrained ResNet18 benchmarking, local inference tools, and a Tkinter accuracy explorer.
// tech stack
PythonPyTorchTorchvisionResNet18TkinterPandas
// preview
