Is it possible to run a Document AI instance in a local container?

[diegolimapereir]

Hi everybody, 

I'm looking for a local solution (ideally using docker) to run some data extraction from some files. Document AI fits perfectly our needs, but we need to run it locally. I have already seen the Deep Learning Containers, but the effort to train models there will be higher than I want.

A Google Cloud competitor has a solution like that, but I want to use the Google Cloud environment. So, in summary, is running some of the features of Document AI in a local container possible?

[McMaco · Accepted solution]

Hi diegolimapereir,

Welcome to the Google Cloud Community!

Google Cloud Document AI currently cannot be installed and run directly on a local container on your machine. Even using Docker, unfortunately, you cannot directly run a Document AI instance locally.

Here's why:

Limited Functionality in Docker: While Docker provides containerization for software, Document AI still relies on:

• Pre-trained Models: These models are not available for local deployment within a Docker container.
• Backend Services: These services are hosted on Google Cloud and are essential for Document AI's functionality.

Focus on Cloud Deployment: Docker is often used for packaging and deploying applications in the cloud. Google designed Document AI as a cloud-based service, leveraging Google's infrastructure for optimal performance and security.

However, Docker can be useful in conjunction with Document AI:

• Containerized Libraries: You can use Docker to manage containerized versions of libraries or tools you use for pre-processing documents before sending them to Document AI (e.g., format conversion, filtering).
• Development and Testing: If you're developing an application that interacts with Document AI, you can use Docker containers to simulate the Document AI environment for testing purposes. This might involve using mock APIs or libraries that mimic Document AI's functionality.
  

Here are some resources that might be helpful:

• Document AI for Extraction: Explore the different Document AI products available on Google Cloud.
• Cloud Vision API with Docker: Learn how to use the Cloud Vision API locally with a Docker container (might be relevant if your needs are more focused on OCR or image analysis).

While you can't directly run Document AI locally with Docker, it can still be a valuable tool in your workflow for development, testing, and pre-processing tasks related to Document AI.

I hope the above information is helpful. Thanks!