Object Detection API

FastAPI service to submit images for object detection and check task status.

Quick Start

• Prerequisites: Python 3.8+, Redis, Google Cloud project (Storage + Pub/Sub enabled)
• Install:

pip install -r requirements.txt

• Configure env (example):

export REDIS_URL=redis://localhost:6379
export GCP_PROJECT_ID=your-gcp-project
export GCS_BUCKET=object-detection-images
export PUBSUB_TOPIC=object-detection-tasks
export TASK_TTL=3600

• Run:

python -m uvicorn src.main:app --host 0.0.0.0 --port 8000

Open http://localhost:8000/docs

Endpoints

• POST /detect: submit an image (multipart field image)

curl -X POST http://localhost:8000/detect \
  -F "image=@/path/to/image.jpg"

• GET /status/{task_id}: check task status and results

curl http://localhost:8000/status/<uuid>

• GET /health: lightweight health (add ?deep=1 for a light GCS probe)
• POST /internal/task-completed: worker callback, returns 204

Environment Variables

• REDIS_URL: Redis connection URL
• GCP_PROJECT_ID: GCP project id
• GCS_BUCKET: GCS bucket for images
• PUBSUB_TOPIC: Pub/Sub topic for tasks
• TASK_TTL: Task TTL in seconds (default 3600)

Project Layout

• src/domain: entities, enums
• src/application: use cases, schemas
• src/infrastructure: providers, Redis/GCS/PubSub
• src/main.py: FastAPI app & routes

Notes

• Detection runs in a separate worker; this API enqueues and tracks tasks
• Blocking SDK calls are wrapped with anyio.to_thread.run_sync to keep the event loop responsive