Releases: deepdoctection/deepdoctection
Releases Β· deepdoctection/deepdoctection
v.0.32.0
What's Changed
- HFLiltTokenClassifier, refactor layoutlm training scripts and re-name cell to sub_layout [extern] by @JaMe76 in #315
- feat: make pdfplumber x-/y-toleration configurable by @FeU-aKlos in #308
- Detectron2 issue 5268 by @JaMe76 in #326
- Sequence classification by @JaMe76 in #331
- Do not disable existing loggers by @chiragjn in #332
- Lazy imports and deprecating Python 3.8 by @JaMe76 in #333
- Lib and device selection by @JaMe76 in #334
New Contributors
- @FeU-aKlos made their first contribution in #308
- @chiragjn made their first contribution in #332
Full Changelog: v.0.31.0...v.0.32.0
v.0.31.0
What's Changed
- Add error module and start refactoring exceptions [utils] by @JaMe76 in #292
- refactor model wrappers [extern][force ci] by @JaMe76 in #294
- Adding Rotation functionality for images by @JaMe76 in #305
- Annotation tracing by @JaMe76 in #309
- Update setup.py by @JaMe76 in #310
- Update setup.py and Github workflows by @JaMe76 in #311
Full Changelog: v.0.30.0...v.0.31.0
v.0.30.0
v.0.29.0
v.0.28.0
What's Changed
- Fixing #233 by @JaMe76 in #236
- Handle relative coords when mapping into Detectron2 structure by @JaMe76 in #237
- Device detection to work with macbooks out of the box. by @aiorga-sherpas in #242
- Refactor
utils
package by @JaMe76 in #248 - Reducing log warnings in text ordering and table refinement process. by @JaMe76 in #252
- Optimize internal data model by @JaMe76 in #253
- Loading 2-dim images with Pillow by @JaMe76 in #259
New Contributors
- @aiorga-sherpas made their first contribution in #242
Full Changelog: v.0.27.0...v.0.28.0
v.0.27.0
What's Changed
- Update opencv by @JaMe76 in #208
- Bumping up opencv-python version to match setup.py by @frivas-at-navteca in #210
- Fix _load_model loading errors when cfg.LIB=="TF" and DocTr is used by @Skifary in #214
- Update doctrocr.py by @JaMe76 in #215
- Adding
AnnotationNmsService
as optional pipeline component by @JaMe76 in #220 - Add additional attribute names by @JaMe76 in #221
- Upgrading DocTr to 0.7 by @JaMe76 in #224
- Make OpenCV optional by @JaMe76 in #229
- Fix #190 and adding Docker CI for building and pushing images to Docker hub by @JaMe76 in #232
New Contributors
Full Changelog: v.0.26...v.0.27.0
v.0.26: Fixing bugs to run with Python 3.10, updating and simplifying third party dependencies and using Torchscript for Detectron2 inference
What's Changed
Full Changelog: v.0.25...v.0.26
v.0.25 Adding more functions and configurations for analyzer
v.0.24 Page methods for generating chunks and W&B integration for LayoutLM training
What's Changed
- Towards llama loader by @JaMe76 in #154
- Adding Layoutlm samples with custom dataset for token classification by @JaMe76 in #156
Full Changelog: v.0.23...v.0.24
v.0.23 Several enhancements for pipeline components and models
What's Changed
- Enhancing text ordering and table segmentation process by @JaMe76 in #134
- Add batch processing between pipeline components via MultiThreadPipelineComponent by @JaMe76 in #137
- Multiple text detectors by @JaMe76 in #138
- Multiple text detectors by @JaMe76 in #139
- Upgrade doctr by @JaMe76 in #144
- Adding HFLayoutLmv3TokenClassifier to the list of token classifiers by @frivas-at-navteca in #141
- Add jdeskew link by @dweekly in #145
- Add documentation on how to build pytorch-cpu docker. by @ryanmcdonough in #147
- Allow loading custom DocTr models in
DoctrTextlineDetector
andDoctrTextRecognizer
by @JaMe76 in #148 - Prioritizing when NMS pairs of categories by @JaMe76 in #149
- Fixing #124 by @JaMe76 in #150
- Adding py.typed by @JaMe76
New Contributors
- @frivas-at-navteca made their first contribution in #141
- @dweekly made their first contribution in #145
- @ryanmcdonough made their first contribution in #147
Full Changelog: v.0.22...v.0.23