Help Needed for Downloading Missing StructureV3 + Layout Models @PaddlePaddle @PaddleOCR

[RajeshGit001]

Hello PaddleOCR Team and Community,

I'm working on a financial document processing project and trying to use PP-StructureV3 for extracting layout and table structures from scanned PDFs. However, I’m unable to download the required models like:

layout_cdla_infer.tar

table_structure_mobile_v2.0_SLANet_infer.tar

ch_PP-OCRv3_rec_infer.tar

These fail with 404 Not Found errors even when tried via the official model_download.sh.

I'd appreciate if anyone could share a working download link or provide the .tar model files if they already have them available. My environment is based on PaddleOCR GPU (2.5.0) in a Linux VM.

Thanks a lot in advance!

Tag: @PaddlePaddle/PaddleOCR 🙏

[liuhongen1234567]

Hello, perhaps you could upgrade the versions of paddle and paddleocr to 3.1.0. This version already supports downloading models from various sources such as Hugging Face, ModelScope, and BOS.

[liuhongen1234567]

Additionally, PaddleOCR 2.5 cannot use PP-StructureV3, and layout_cdla_infer.tar seems to be an earlier model. The models table_structure_mobile_v2.0_SLANet_infer.tar and ch_PP-OCRv3_rec_infer.tar were officially provided and exported based on Paddle 3.0. These models can be found in the PP-StructureV3 documentation https://www.paddleocr.ai/latest/en/version3.x/pipeline_usage/PP-StructureV3.html#1-introduction-to-pp-structurev3-pipeline model list, but they only support Paddle 3.0.

[zhangyubo0722]

It is recommended to check the latest documentation at https://github.com/PaddlePaddle/PaddleOCR/blob/main/docs/version3.x/pipeline_usage/PP-StructureV3.en.md