Enhance SageMaker Notebooks To Use YOLO Models

[rajjainl]

This pull request was generated by @kiro-agent 👻

Comment with /kiro fix to address specific feedback or /kiro all to address everything.
_{Learn about Kiro autonomous agent}

---

Summary

This PR adds a new SageMaker notebook that uses YOLO models for defect detection as an alternative to the existing Lookout for Vision approach.

Changes

New File: DDA_SageMaker_YOLO_Training.ipynb

A comprehensive notebook for training YOLOv8 models on SageMaker:

• Step 1-2: Environment and S3 setup
• Step 3: Data preparation - converts existing cookie dataset to YOLO format
• Step 4: Training script creation using Ultralytics YOLOv8
• Step 5: SageMaker PyTorch estimator configuration and training job launch
• Step 6-7: Model download and optional SageMaker Neo compilation

Advantages of YOLO over Lookout for Vision:

• ✅ Open-source - No AWS Marketplace subscription required
• ✅ Flexible - Supports detection, segmentation, and classification
• ✅ Fast - Optimized for real-time edge inference
• ✅ Portable - Exports to ONNX and TorchScript formats

Updated: README.md

• Added new "SageMaker Training Notebooks" section documenting both options
• Updated deployment instructions to reference both notebook options

Testing

The notebook is designed to be run in Amazon SageMaker Notebook Instance or SageMaker Studio. It uses the same cookie dataset as the existing LFV notebooks for consistent testing.

Deployment Compatibility

YOLO models can be deployed using the same DDA edge deployment workflow:

1. Export model to ONNX/TorchScript
2. Package as Greengrass component
3. Deploy via DDA_Greengrass_Component_Creator.ipynb