Skip to content

README.md

Latest commit

 

History

History
42 lines (30 loc) · 1.01 KB

README.md

File metadata and controls

42 lines (30 loc) · 1.01 KB

Building Classification Example

End-to-end finetuning of resnet18-classification on Banepa Municipality, Nepal OAM imagery with binary building/no_building labels derived from OSM segmentation data.

Prerequisites

  • zenml
  • Sample data in data/sample/ (OAM tiles + OSM labels, including pre-generated classification_labels.csv)

Quick Start

uv sync --group example --group local
just setup
uv run python examples/classification/run.py

Workflow

The script runs the full workflow in one execution:

  1. Initialize ZenML and local STAC context
  2. Register the base model item
  3. Register the dataset item
  4. Finetune the model
  5. Promote the finetuned model
  6. Run prediction on sample imagery

CI Usage

FAIR_FORCE_CPU=1 uv run python examples/classification/run.py

Output

Artifact Location
STAC items stac_catalog/
Trained artifacts artifacts/
Predictions data/sample/predict/predictions/*.csv