This project demonstrates a full-stack AI solution with:

• Backend: FastAPI service providing an /infer endpoint for object detection using YOLOv7-tiny.
• Frontend: Streamlit app to upload images and visualize inference results.
• Deployment: Both services containerized with Docker and orchestrated via Docker Compose.

• Python 3
• FastAPI (backend REST API)
• Streamlit (frontend UI)
• YOLOv7-tiny (pretrained AI model)
• Docker & Docker Compose

Projectrestapi/

• │── backend-aiservice/
• │ ├── app.py
• │ ├── requirements.txt
• │ ├── Dockerfile
• │ └── yolov7-tiny.pt
• │
• │── frontend-uiservice/
• │ ├── app.py
• │ ├── requirements.txt
• │ └── Dockerfile
• │
• │── docker-compose.yaml

• Backend (FastAPI):

  • Implements /infer endpoint for image inference.
  • Loads YOLOv7-tiny model and returns detections.
  • Run: uvicorn app:app --host 0.0.0.0 --port 8000

• Frontend (Streamlit):

  • UI with file uploader.
  • Sends images to backend /infer endpoint.
  • Displays results.
  • Runs on port 8501.

• Docker & Compose:

  • Each service has its own Dockerfile.
  • docker-compose.yaml defines services and network.
  • Frontend connects to backend using AI_BASE_URL env var.

-Prerequisites:

• Install Docker and Docker Compose.

• Steps:
  1. Open terminal in project directory.
  2. Run: docker-compose up --build
  3. Access:
     • Backend API: http://localhost:8000/docs
     • Frontend UI: http://localhost:8501

• Optimizing model with onnx or TRT
• Creating own yolov7 class Rather than using yolov7 github
• Optimizing UI define for multiple images

• FastAPI Docs: https://fastapi.tiangolo.com/
• Streamlit Docs: https://docs.streamlit.io/
• YOLOv7: https://github.com/WongKinYiu/yolov7
• Docker Docs: https://docs.docker.com/
• Docker Compose: https://docs.docker.com/compose/