deepdoctection
deepdoctection is the main package for running and training models. It provides the pipeline framework, model wrappers, built-in pipelines, training scripts and evaluation methods.
The base package only installs the necessary dependencies for running inference with some selected models. For training, evaluating as well as running all available models, the full package needs to be installed.
Overview
- analyzer: Configuration and factory functions for creating document analysis pipelines and the built-in analyzer.
- configs: YAML configuration for pipelines and model profiles for the model catalogue.
- extern: External model wrappers (Detectron2, DocTr, HuggingFace Transformers, Tesseract, PdfPlumber, etc.)
- pipe: Pipeline components and services.
- eval: Evaluation metrics and Evaluator.
- train: Training utilities and training scripts for Detectron2 and selected Transformer models.
Installation
Basic Installation
For inference use cases, install the base package:
uv pip install deepdoctection
Important: Various dependencies must be installed separately:
- PyTorch: Follow instructions at https://pytorch.org/get-started/locally/ according to your os and hardware.
- Transformers:
pip install transformers>=4.48.0(if using HF models) - Timm:
pip install timm>=0.9.16(necessary for if using some dedicated HF models) - DocTr:
pip install python-doctr>=1.0.0(if using DocTr models) - Detectron2: Follow instructions at https://detectron2.readthedocs.io/en/latest/tutorials/install.html
- PDFPlumber:
pip install pdfplumber>=0.11.0 - JDeskew:
pip install jdeskew>=0.2.2 - Boto3:
pip install boto3==1.34.102
For running evaluation with various metrics you can also install in then use:
- APTED:
pip install apted==1.0.3 - Distance:
pip install distance==0.1.3 - Pycocotools:
pip install pycocotools>=2.0.2
Image processing is supported by PIL or OpenCV. PIL is used by default and will always be installed. If you prefer to use OpenCV, you can install it:
- OpenCV:
pip install opencv-python==4.8.0.76
Full Installation (Training & Evaluation)
For a one large install with all dependencies (except PyTorch), run:
uv pip install deepdoctection[full]
Development Installation
For development purpose use clone the repository and install in editable mode.
License
Apache License 2.0
Author
Dr. Janis Meyer
1 maintainer
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file deepdoctection-1.2.14.tar.gz.
File metadata
- Download URL: deepdoctection-1.2.14.tar.gz
- Upload date:
- Size: 151.5 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/7.0.0 CPython/3.10.20
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
deb3feabceff00f2d11886caebe42f8ea37e846e55d91545e779cd568f4690e8
|
|
| MD5 |
622e313725b237e6dd3b4967da17e0eb
|
|
| BLAKE2b-256 |
fd076dc253e2709599860b666e5a0ebd6ae8eb76cece4aa0d5a5076e707e96ed
|
File details
Details for the file deepdoctection-1.2.14-py3-none-any.whl.
File metadata
- Download URL: deepdoctection-1.2.14-py3-none-any.whl
- Upload date:
- Size: 192.6 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/7.0.0 CPython/3.10.20
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
87ea827fb381a8a4c0dcaa7a2aa42075d8091ad24ec91afefd511cb5e801dde0
|
|
| MD5 |
4fc0da2504d46223c43cf522eaacbc82
|
|
| BLAKE2b-256 |
f33653c1df9b2b0a922e0904684c4b966e0e47718239eda8c5adf10e7630661f
|
