marver VS paperless-ngx

marver https://github.com/sylv/marver

My NAS has a load of files dumped onto it over the years. Some are organised neatly, some are in a dumping ground that I promised myself I'd sort through and never did. Some are in old folders that I haven't touched in years. Instead of spending a couple weekends sorting through it like a sane person, I'm making something to index it and hopefully replace Plex in the process.

The goal for marver is to index all those files and rip as much metadata as possible from each, then display that in a pretty UI and give the user multiple ways to refine the metadata, search through their files and make the most of the piled up 1s and 0s they have.

So far I've mostly just been having fun making individual pieces - image search powered by CLIP, extracting information from hard to parse file names using llama.cpp, pulling information from EXIF on images, some building blocks for intro detection on videos. Right now I'm porting the machine learning code to rust so I can bundle it all into a single, small docker container a lot easier, and to learn how the black boxes I'm using actually work, at least on a basic level.

I've tried similar apps but they all either have mediocre UI, are painful to setup, or make me do all the indexing manually. I want something that most people can deploy in a few minutes and that can show off a lot of what it can do immediately, while still letting the user refine the results from automated approaches. The UI has to be amazing to bundle this all into something that is actually useful, but it's been really fun putting it all together, piece by piece.

It's still a bit all over the place, but my NAS has annoyed me enough recently that I'm going to be focusing on getting a version that can be deployed, even if it's missing most of the features I want, so that it can start being useful to me. I think once that happens I'll be more motivated to add features, especially with a solid base I can work from.

I steered a friend towards Paperless (and away from an LLM solution) as a way of searching/accessing GBs of architectural PDFs recently - so far, it’s apparently working well for them.

https://github.com/paperless-ngx/paperless-ngx

Paperless-ngx is a document management system that transforms your physical documents into a searchable online archive so you can find your physical documents easier. With features such as tags, full text search, multi-user permissions system, this is a dream for those who like to keep an organized folder of files and documents.

Since last year I’ve been configuring and maintaining my homelab setup and it is just amazing.

I’ve learned so much about containers, virtual machines and networking. Some of the self hosted applications like paperless-ngx [1] and immich [2] are much superior in terms of features than the proprietary cloud solutions.

With the addition of VPN services like tailscale [3] now I can access my homelab from anywhere in the world.

The only thing missing is to setup a low powered machine like NUC or any mini PC so I can offload the services I need 24/7 and save electricity costs.

If you can maintain it and have enough energy on weekends to perform routine maintenance and upgrades. I would 100% recommend setting up your own homelab.

[1] https://docs.paperless-ngx.com/

This has been posted a few times already, but I cannot tell you how life changing Paperless NGX is for organizing PDFs. As someone who wrangles all of the insurance and bills for my house, this open source software is so damn good.

https://docs.paperless-ngx.com/

I maintain Bash script to quickly set it up locally on Linux with Podman. Give it a spin if you want to kick the tires.

https://github.com/jdoss/ppngx

Personally, I use https://github.com/joeyates/imap-backup to archive all my emails and then only keep them on the remote server for as long as I need to (basically until I read them and respond or download an attachment into https://docs.paperless-ngx.com )