sh4den/Montscan

What it solves

Montscan automates the tedious process of managing scanned documents. Instead of manually renaming files like "scan001.pdf" and moving them to the correct folders, it uses Vision AI to analyze the content of the scan and automatically generate descriptive, date-aware filenames.

How it works

The system acts as a bridge between a network scanner and a storage destination. It uses three possible "ingress" methods to receive files: an integrated FTP server, a Samba (SMB) poller, or a local folder watcher. Once a PDF is received, the tool converts the PDF to an image and sends it to a local Ollama instance running a vision-capable model (such as LLaVA or Llama 3.2-Vision). The AI analyzes the image to suggest a meaningful name, and the processed file is then sent to an "egress" provider, such as a WebDAV server (Nextcloud/ownCloud) or a Samba share.

Who it’s for

It is designed for users who have network scanners and want to automate their digital archiving workflow without relying on cloud-based AI services, keeping the processing local via Ollama.

Highlights

• Local Vision AI: Uses Ollama for private, local document analysis and naming.
• Flexible Ingress: Supports FTP, SMB polling, and local folder monitoring.
• Multiple Egress Options: Can upload to WebDAV servers or network shares (SMB/CIFS).
• Docker Ready: Includes Docker Compose support with optional GPU acceleration for faster AI processing.