microsoft / unilm

Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities
https://aka.ms/GeneralAI
MIT License
20.19k stars 2.55k forks source link

Inference script on DiT object detection does not work #1345

Open domaschh opened 1 year ago

domaschh commented 1 year ago

Describe the bug Model I am using (UniLM, MiniLM, LayoutLM ...): DiT classification

The problem arises when using:

I set up the environment like described but when I try to use the inference.py like described in the documentation I get Failed to download https://layoutlm.blob.core.windows.net/dit/dit-fts/publaynet_dit-b_mrcnn.pth urllib.error.HTTPError: HTTP Error 409: Public access is not permitted on this storage account

To Reproduce Steps to reproduce the behavior:

  1. Setup environemtn for DiT
  2. (https://github.com/microsoft/unilm/tree/master/dit/object_detection)

Expected behavior A clear and concise description of what you expected to happen. Expect to be able to download the weights.

### Tasks
sugia commented 8 months ago

A workable solution is to download the weights as a zip file: https://layoutlm.blob.core.windows.net/dit/dit-fts/publaynet_dit-b_mrcnn.pth?sv=2022-11-02&ss=b&srt=o&sp=r&se=2033-06-08T16:48:15Z&st=2023-06-08T08:48:15Z&spr=https&sig=a9VXrihTzbWyVfaIDlIT1Z0FoR1073VB0RLQUMuudD4%3D

Then points model weights to that zip file:

python ./dit/object_detection/inference.py
--image_path ./dit/object_detection/publaynet_example.jpeg
--output_file_name output.jpg
--config ./dit/object_detection/publaynet_configs/maskrcnn/maskrcnn_dit_base.yaml
--opts MODEL.WEIGHTS publaynet_dit-b_mrcnn.zip

More details can be found here: https://github.com/microsoft/unilm/issues/1246

Update: The zip file link above is no longer working. I have a hunch that the DIT team doesn't really want anyone to get access to their pre-train weights.

jesusfbes commented 7 months ago

Hello, I have trying this kind of workaround and keep getting authentication errors. Can I find the weights anywhere else?

jnm-ronquillo commented 5 months ago

I have found the weights here: https://huggingface.co/Sebas6k/DiT_weights/tree/main from this example code: https://huggingface.co/spaces/nielsr/dit-document-layout-analysis/blob/main/app.py