How nanonets/signaturedetectiondataset works
Project Type and Purpose
• This is a Python-based sample client for NanoNets Object Detection.
• Purpose: demonstrate end-to-end workflow for training and using a signature-detection model via NanoNets REST API.
• It is not a standalone backend or frontend web app, but a CLI tool + local data assets driving a cloud ML API.
File Structure
• Top-level: README.md, draw_boxes.py, annotated/ (visualized images), images/ (raw inputs), annotations/ (JSON & XML payloads).
• code/: Python scripts implementing each stage:
– create-model.py (calls POST /model)
– upload-training.py (uploads images+annotations)
– train-model.py (fires training job)
– model-state.py (polls job status)
– prediction.py (requests inference)
• Dependencies: requests library; uses environment variables NANONETS_API_KEY and NANONETS_MODEL_ID.
README for Additional Insights
• Workflow steps correspond one-to-one to the scripts.
• Uses converted JSON payloads for convenience (no XML parsing).
• All network interactions go through NanoNets REST endpoints.