Install with Docker

Install with Docker

The Docker setup runs Docveta, its database and text recognition together, on any machine with Docker (x64 or ARM 64-bit).

Start

mkdir docveta && cd docveta
curl -LO https://raw.githubusercontent.com/anand34577/docveta/main/deploy/docker-compose.yml
docker compose up -d

Open http://<server>:8080 and create your account. There is nothing to edit:

• the database password is generated on first start and kept in a Docker volume that only the database and Docveta can read
• the text recognition engine (PaddleOCR on the processor) fetches its own token from Docveta with an enrollment key that only the engine containers can see
• the secret key that protects stored passwords is generated and saved in Docveta's data

Your data lives in the Docker volumes docveta_data (documents, and docveta.conf with the secret key) and docveta_db (the database); docveta_secrets holds the database password.

More engines

docker compose --profile tesseract up -d      # Tesseract: 100+ languages
docker compose --profile npu-rockchip up -d   # Rockchip NPU boards (RK3588/RK3576/RK3566)

The Rockchip engine needs models converted for your board, see OCR engines. For a GPU, run the docveta-ocr package on the host or a PC with a GPU and point it at Docveta, see OCR engines.

Settings

Optional settings (port, the public address behind a reverse proxy, …) go in a .env file next to docker-compose.yml; see .env.example. Any setting from Configuration can also be added to the environment: section of the docveta service.

Everyday commands

docker compose logs -f docveta                     # log
docker compose pull && docker compose up -d        # upgrade
docker compose down                                # stop (data stays in the volumes)
docker compose exec docveta /docveta doctor        # health check

Backups: back up the docveta_data, docveta_db and docveta_secrets volumes together, for example with docker compose stop and a copy of /var/lib/docker/volumes/docveta_*. Keep docveta_secrets with the database: without it a restored docveta_db gets a new password on start that the database doesn't know, and Docveta can't connect. Or dump the database as in Backup and restore.

Upgrading from an older compose file

Installs that set POSTGRES_PASSWORD and folder paths (DOCVETA_DATA_DIR=./data) in .env keep working unchanged: the new compose file reuses that password and those folders. Worker tokens in .env are still used; remove them to let the engines enroll themselves.

Images

Image	Contents
ghcr.io/anand34577/docveta	Docveta (amd64, arm64)
ghcr.io/anand34577/docveta-ocr	PaddleOCR engine, CPU (amd64, arm64)
ghcr.io/anand34577/docveta-worker-tesseract	Tesseract engine (amd64, arm64)
ghcr.io/anand34577/docveta-worker-rknn	Rockchip NPU engine (arm64)