Zoe Depth Estimation

This project utilizes the ZoeDepth model to generate depth map predictions for a given image. It includes a Gradio web interface, a FastAPI-based API, and a command-line interface (CLI).

This project is based on this repository by Mert Çobanov and was implemented for educational purposes.

Features

• Gradio Interface: An interactive web demo to easily upload images and visualize their depth maps.
• FastAPI: A RESTful API server for performing depth estimation.
• CLI: A command-line tool for quick processing of single images.
• Poetry: Modern and reliable dependency management.
• Docker Support: For easily packaging and running the application in an isolated environment.
• Jenkins Integration: A CI/CD workflow to automate code quality checks and build processes.

Requirements

Before you begin, ensure you have the following tools installed on your system:

• Python 3.12+
• Poetry
• Docker (if you plan to run the application with Docker)

Installation and Setup

1. Clone the Repository:

   git clone github.com/gokaymeydan/depth-estimation
   cd depth-estimation

2. Configure Environment Variables: The project requires an IMAGE_API_KEY to upload the resulting depth maps. Create a file named .env in the project's root directory and add your API key from ImgBB.

   .env file content:

   IMAGE_API_KEY="YOUR_API_KEY_HERE"
   

3. Install Dependencies: Poetry will install all project dependencies into a virtual environment.

   poetry install

Usage

1. Local Development (with Poetry)

Use this method for development and testing the application locally.

• To launch the Gradio Web Interface:

  poetry run python gradio_app.py

  Then, open http://127.0.0.1:7860 in your browser.

• To launch the FastAPI Server:

  poetry run uvicorn api:app --host 0.0.0.0 --port 8041

  The API will be accessible at http://127.0.0.1:8041/docs.

• To use the Command-Line Interface (CLI):

  poetry run python cli.py <path_to_input_image> <path_to_output_image>

  Example:

  poetry run python cli.py input.jpg output.png

2. Running with Docker (Deployment)

Use this method to run the application in an isolated container.

1. Build the Docker Image:

   docker build -t depth-estimation .

2. Run the Docker Container: This command starts the FastAPI server, using the API key from your .env file.

   docker run -d -p 8041:8041 --env-file .env --name depth-api depth-estimation

   The application will now be running at http://localhost:8041.

CI/CD

This project includes a Jenkinsfile for automation. This workflow automatically performs the following steps on each code push:

• Checks code quality with black and isort.
• Builds the project's Docker image.

License

This project is licensed under the MIT License. See the LICENSE file for more details.