Trying to run it... but it will not work
Hi! This is amazing exacly what we are looking for. But unfortunally we cannot make it work.
After the docker is build and making the request there is never an answer. But in the docker desktop logs we see some respones from the service, but it is not very sure what everything is.
Would it be possible for you to write a bit more documentation and maybe 1 example on how to run it?
With kind regards,
Lennart
Hi,
The whole process is actually running in 3 steps:
1- Start the service
// If you have GPU in your system:
make start
// Without GPU support:
make start_no_gpu
2 - Get the segments:
// If you want to use the visual models [which require more resources] (Please do not forget to put the correct path of your pdf file.)
curl -X POST -F 'file=@/PATH/TO/PDF/pdf_name.pdf' localhost:5060
// If you want to use the faster - non-visual models:
curl -X POST -F 'file=@/PATH/TO/PDF/pdf_name.pdf' localhost:5060/fast
3 - Stop the service:
make stop
If this does not solve your problem please give us more information about what you are experiencing, maybe with some screenshots.
Also, we have shared some pdf files to test the service - if it's working, you can find them in project repo and try them: https://github.com/huridocs/pdf-document-layout-analysis/tree/main/test_pdfs
Please let us know the outcomes.
Best,
Ali
Hi, Ali
Thank you for the quick answer. I tried this 3 step apporach, but i just do not get a repsonse by the curl. Maybe it is a compute problem.
EDIT: The /fast enpoint works! Thanks. The normal endpoint can not be used by my laptop. That was it
Hi Lennart,
You're welcome. When there is not enough GPU memory in your system, the process switches to use CPU and running visual models with CPU is taking quite some time - especially if the PDF file is a big one.
I'm glad that it's working, thanks for your interest again!