💎 YOLOv8 Image Inference via AWS SageMaker, Lambda, and API Gateway

This project enables image object detection using a YOLOv8 model hosted on AWS SageMaker, accessible via a REST API powered by Lambda and API Gateway. Users can send images via a REST API endpoint and receive annotated images with detected bounding boxes.

---

📚 Documentation

All setup instructions, deployment details, and troubleshooting steps are organized in the docs/ folder.

📄 Document	Description
1-overview.md	System architecture and high-level project overview
2-sagemaker-setup.md	Setup SageMaker model, inference script, and deployment
3-lambda-function.md	Lambda function logic, permissions, and deployment
4-api-gateway.md	API Gateway configuration, routes, and URL structure
5-local-testing.md	Test from local PC using curl or Python requests
6-troubleshooting.md	Fix common issues, error codes, and log locations
7-references.md	Useful external links and AWS documentation

---

🚀 Local testing

mim download mmyolo \
  --config yolov8_x_mask-refine_syncbn_fast_8xb16-500e_coco \
  --dest models/yolov8_x_mask-refine_syncbn_fast_8xb16-500e_coco

Installation (CPU)

conda create -n mmyolo python=3.8
conda activate mmyolo
pip install -r src/requirements.txt
pip install -r scripts/requirements.txt

Deployment to SageMaker

AWS_PROFILE=your-profile python scripts/deploy.py

Inference

python src/test_yolo_locally.py # Should show a window with detection