Skip to content
Snippets Groups Projects

Compare revisions

Changes are shown as if the source revision was being merged into the target revision. Learn more about comparing revisions.

Source

Select target project
No results found
Select Git revision
  • bump-boto3
  • bump-wandb
  • bump-nltk
  • bump-pandas
  • bump-PyYAML
  • bump-lxml
  • bump-mlflow-skinny
  • bump-scipy
  • main
  • bump-numpy
  • bump-matplotlib
  • bump-torchvision
  • bump-torchaudio
  • bump-torch
  • bump-tensorboard
  • bump-mdutils
  • bump-mkdocs-material
  • bump-mkdocstrings-python
  • bump-albumentations
  • support-nested
  • bump-griffe
  • half_precision
  • unit-test-polygon-bug
  • bump-black
  • bump-imageio
  • bump-teklia-nerval
  • bump-tqdm
  • configs-fix-eval
  • memory-snapshot
  • 308-support-rotation-on-dan-dataset-extraction
  • remove-requirements-files
  • memory-profile
  • use-kornia
  • visualisation-predictions-validation-steps
  • move-unknown-token-replace
  • visualize-transcription-inference-gif
  • tmp
  • objects-to-plot-attention-map
  • 278-padding-tokens-are-not-ignored-during-training
  • fix-attention_vis
  • 281-add-support-for-prompting
  • bump-mkdocstrings
  • bump-mkdocs
  • model-compile-suppress-errors
  • bump-sentencepiece
  • memory-leak
  • merge-datasets
  • multi-dan
  • display-pred
  • bump-prettytable
  • memory-profiler
  • memleak
  • multi-dan-perceiver
  • lower-memory-usage
  • configuration-file-for-training
  • try-out-pytest-cov
  • parse-labels-json
  • multi-output
  • perceiver
  • label-smoothing
  • multi-output-old
  • fix-img-position
  • bump-opencv-python
  • test-dropout-dan
  • dan-temperature-scaling
  • split-charset
  • merge-training-managers
  • bump-fontTools
  • remove-syn-lines
  • rename-apply-teacher-forcing-to-add-label-noise-in-training-code
  • prompt-socface
  • bump-mlflow
  • combine-with-bert
  • train-random-start-token
  • 35-new-function-to-aggregate-attention-maps-2
  • 35-new-function-to-aggregate-attention-maps
  • fix-gifs
  • revert-changes-to-config-and-fix-publication
  • update-training-doc
  • 0.1.0
  • 0.2.0
  • 0.2.0-dev1
  • 0.2.0-dev2
  • 0.2.0-dev3
  • 0.2.0-dev4
  • 0.2.0-dev5
  • 0.2.0-dev6
  • 0.2.0rc1
  • 0.2.0rc10
  • 0.2.0rc11
  • 0.2.0rc12
  • 0.2.0rc2
  • 0.2.0rc3
  • 0.2.0rc4
  • 0.2.0rc5
  • 0.2.0rc6
  • 0.2.0rc7
  • 0.2.0rc8
  • 0.2.0rc9
  • 0.2.1
  • 0.2.2a1
  • 0.2.2a2
  • 0.2.2b1
  • 0.2.2b2
  • 0.2.2rc1
  • 0.2.2rc2
  • 0.2.2rc3
  • 0.2.2rc4
  • attention-exp
109 results

Target

Select target project
No results found
Select Git revision
  • bump-boto3
  • bump-wandb
  • bump-nltk
  • bump-pandas
  • bump-PyYAML
  • bump-lxml
  • bump-mlflow-skinny
  • bump-scipy
  • main
  • bump-numpy
  • bump-matplotlib
  • bump-torchvision
  • bump-torchaudio
  • bump-torch
  • bump-tensorboard
  • bump-mdutils
  • bump-mkdocs-material
  • bump-mkdocstrings-python
  • bump-albumentations
  • support-nested
  • bump-griffe
  • half_precision
  • unit-test-polygon-bug
  • bump-black
  • bump-imageio
  • bump-teklia-nerval
  • bump-tqdm
  • configs-fix-eval
  • memory-snapshot
  • 308-support-rotation-on-dan-dataset-extraction
  • remove-requirements-files
  • memory-profile
  • use-kornia
  • visualisation-predictions-validation-steps
  • move-unknown-token-replace
  • visualize-transcription-inference-gif
  • tmp
  • objects-to-plot-attention-map
  • 278-padding-tokens-are-not-ignored-during-training
  • fix-attention_vis
  • 281-add-support-for-prompting
  • bump-mkdocstrings
  • bump-mkdocs
  • model-compile-suppress-errors
  • bump-sentencepiece
  • memory-leak
  • merge-datasets
  • multi-dan
  • display-pred
  • bump-prettytable
  • memory-profiler
  • memleak
  • multi-dan-perceiver
  • lower-memory-usage
  • configuration-file-for-training
  • try-out-pytest-cov
  • parse-labels-json
  • multi-output
  • perceiver
  • label-smoothing
  • multi-output-old
  • fix-img-position
  • bump-opencv-python
  • test-dropout-dan
  • dan-temperature-scaling
  • split-charset
  • merge-training-managers
  • bump-fontTools
  • remove-syn-lines
  • rename-apply-teacher-forcing-to-add-label-noise-in-training-code
  • prompt-socface
  • bump-mlflow
  • combine-with-bert
  • train-random-start-token
  • 35-new-function-to-aggregate-attention-maps-2
  • 35-new-function-to-aggregate-attention-maps
  • fix-gifs
  • revert-changes-to-config-and-fix-publication
  • update-training-doc
  • 0.1.0
  • 0.2.0
  • 0.2.0-dev1
  • 0.2.0-dev2
  • 0.2.0-dev3
  • 0.2.0-dev4
  • 0.2.0-dev5
  • 0.2.0-dev6
  • 0.2.0rc1
  • 0.2.0rc10
  • 0.2.0rc11
  • 0.2.0rc12
  • 0.2.0rc2
  • 0.2.0rc3
  • 0.2.0rc4
  • 0.2.0rc5
  • 0.2.0rc6
  • 0.2.0rc7
  • 0.2.0rc8
  • 0.2.0rc9
  • 0.2.1
  • 0.2.2a1
  • 0.2.2a2
  • 0.2.2b1
  • 0.2.2b2
  • 0.2.2rc1
  • 0.2.2rc2
  • 0.2.2rc3
  • 0.2.2rc4
  • attention-exp
109 results
Show changes

Commits on Source 4

2 files
+ 6
− 4
Compare changes
  • Side-by-side
  • Inline

Files

dan/ocr/predict/attention.py

+5 −3
Original line number Diff line number Diff line
@@ -219,15 +219,17 @@ def get_predicted_polygons_with_confidence( 
            max_object_height=max_object_height,

            size=(resized_sizes[1], resized_sizes[0]),

        )

        start_index += len(text_piece) + offset

        if not polygon:

            continue



        # Apply image resizing to polygon

        polygon["polygon"] = resize_polygon(

            polygon=polygon["polygon"],

            original_image_size=original_sizes,

            ratios=ratios,

        )

        start_index += len(text_piece) + offset

        if not polygon:

            continue



        polygon["text"] = text_piece

        polygon["text_confidence"] = confidence

        polygons.append(polygon)

pyproject.toml

+1 −1
Original line number Diff line number Diff line
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta" 


[project]

name = "atr-dan"

version = "0.2.2a1"

version = "0.2.2b1"

description = "Teklia DAN"

readme = "README.md"

requires-python = ">=3.10"