Skip to content

Latest commit

 

History

History
818 lines (644 loc) · 32.5 KB

CHANGELOG.md

File metadata and controls

818 lines (644 loc) · 32.5 KB
Raw

Changelog

Python versions

The following table shows which versions of MapReader are compatible with which versions of python:

Python MapReader
3.9+ v1.2.0
3.8-3.11 <= v.1.1.5

Pre-release

Add new changes here

Fixed

  • Fixes the explore_predictions and explore_search_results methods in the DeepSoloRunner and MapTextRunner classes (#544)

v1.6.2 (2025-01-02)

Added

  • Added ability to save and reload text predictions (#536
  • Added minimal dataclasses for text predictions (#536
  • Added skip_blank_patches argument to MapImages.patchify_all() (#540)

v1.6.1 (2024-11-18)

Added

  • Added infer_parent_coords_from_patches method to MapImages class (#511)
  • Added explore_patches method to MapImages class for interactive exploration of patches (#511)
  • Added explore_predictions and explore_search_results methods to text runners (#511)

Changed

  • Metrics are now stored in a nested dictionary (e.g. metrics['train']['loss'] instead of metrics["epoch_loss_train]) (#511)
  • plot_metric method is simplified - see updated docs (#511)
  • Renamed show_parent as show_patches (#511)
  • Default value for add_id argument when plotting metadata in SheetDownloader is now False (#529)

v1.5.1 (2024-10-30)

Added

  • Added text-requirement.txt file to install text spotting dependencies. Users can now install text spotting dependencies using pip install -r text-requirements.txt (#515)

Removed

  • Removed text extras from setup.py (#515)

Fixed

  • Fixes upload to PyPI by removing text extras (#515)

v1.5.0 (2024-10-29)

Added

  • Ability to install all three text-spotting frameworks at same time (#514)
  • Can install text spotting dependencies using pip install mapreader[text] (#514)
  • Text spotting code now covered by tests (#514)

v1.4.1 (2024-09-17)

Changed

  • Use tqdm.auto for progress bars

v1.4.0 (2024-09-12)

Added

  • check_georeferencing method and georeferenced attribute added to MapImages class (#495)
  • MapImages.convert_images method now supports saving to GeoJSON format (set save_format="geojson") (#495)
  • All file loading methods now support pathlib.Path and gpd.GeoDataFrame objects as input (#495)
  • Loading of dataframes from GeoJSON files now supported in many file loading methods (e.g. add_metadata, Annotator.__init__, AnnotationsLoader.load, etc.) (#495)
  • load_frames.py added to mapreader.utils. This has functions for loading from various file formats (e.g. CSV, Excel, GeoJSON, etc.) and converting to GeoDataFrames (#495)
  • Added tests for text spotting code (#500)
  • Added search_preds, show_search_results and save_search_results_to_geojson methods to text spotting code (#502)

Changed

  • Refactoring of SheetDownloader to make full use of geopandas functionality - "metadata.json" is now read in as a GeoDataFrame instead of json dictionary (#495)
  • query_map_sheets_by_string and download_map_sheets_by_string methods now search using columns in the GeoDataFrame instead of keys in the json dictionary (#495)
  • columns argument renamed to usecols in MapImages.add_metadata method (to align with pandas) (#495)
  • polygon column renamed to geometry (to align with geopandas) (#495)

Removed

  • hist_published_dates method removed from SheetDownloader as it is no longer needed. Use sd.metadata["published_date"].hist() instead (#495)

v.1.3.10 (2024-09-04)

No changes to code. This release marks the move the the maps-as-data github organisation.

v1.3.9 (2024-08-21)

Added

  • Changelog added (#461)

Changes for versions <= v1.3.8 were added retroactively and may not be complete.

v1.3.8 (2024-08-12)

Fixed

  • Fixes the get_label_index method in AnnotationsLoader class (#490)

v1.3.7 (2024-08-12)

Added

  • Adds option to show a border around central patch when annotating with context (#480)
  • Adds show_vals argument to show patch information when annotating (#480)
  • Adds ability to specify labels map when loading annotations (#489)

v1.3.6 (2024-08-09)

Added

  • Adds new framework (MapTextPipeline) for running text spotting (#474)
  • Adds occlusion analysis to MapReader post processing (#470)

Changed

  • Better error messaging #476

Fixed

v1.3.5 (2024-08-05)

Added

  • Adds sphinx-autobuild for live updating of documentation (#457)
  • Adds data warning when downloading large amounts of data (#464)

v1.3.4 (2024-07-05)

Added

  • Adds deduplication method to text spotting code #435
  • Allows users to create overlapping patches #435

v1.3.3 (2024-07-03)

Changed

  • Updates workshop notebooks (#424)

v1.3.2 (2024-05-17)

Added

  • Adds CLI command to check installation and print MapReader version (#431)
  • Adds calculation of average across all channels for mean/std pixel stats (#434)

Changed

  • Updates to worked examples and corresponding READMEs (#414)

Fixed

v1.3.1 (2024-05-03)

Fixed

v1.3.0 (2024-04-29)

This release was used for the Data/Culture Spring Workshop 2024.

Added

  • Adds text spotting to MapReader (#388)

v1.2.0 (2024-04-08)

This release removes support for Python 3.8.

Added

  • Support for Python 3.12 (#384

Removed

  • Support for Python 3.8 (#384)

v1.1.5 (2024-04-04)

Added

  • Adds download_in_parallel argument to download methods (#363)

Removed

  • Conda install option (#366)

Fixed

  • Re-adds square_cuts argument to patchify images with deprecation warning (#373)
  • Fixes functions which reproject coordinates when converting between coordinate systems (#374)

v1.1.4 (2024-02-23)

Fixed

  • Annotator no longer errors if "url" not in dataframe (#357)

v1.1.3 (2024-02-21)

Added

  • Adds context-based post-processing script (#342)
  • Adds save_predictions method to the ClassifierContainer class (#342)
  • Adds filter_for argument to the Annotator class (#319)
  • Adds ability to create a PatchContextDataset from the AnnotationsLoader class (#350)

Changed

  • The ClassifierContainer class now works with both PatchDatasets and PatchContextDatasets (#350)
  • Context classifier now works using a single branch model, one image, containing the patch and its context are used as input (#350)

Removed

  • Removes square_cuts option when patchifying maps (#350)

v1.1.2 (2024-02-05)

Fixed

v1.1.1 (2024-01-08)

Added

  • Adds add_coords_from_grid_bb method to the MapImages class, allowing users to regenerate coordinates from grid bounding boxes (#318)
  • Adds save_parents_as_geotiffs method to the MapImages class, enabling users to save parent images as GeoTIFFs (#318)
  • Ensures CSV files are loaded correctly by applying literal_eval to all columns in the dataframe (#318)
  • Adds first draft of JOSS paper (#321)

Changed

  • Annotator no longer requires URL column in parent dataframe (#337)
  • Annotator no longer accepts arbitrary kwargs (#337)

Fixed

v1.1.0 (2023-12-14)

Fixed

  • Fixes the reproject_geo_info method in geo_utils (#317)

v1.0.7 (2023-12-14)

New annotation interface!

Added

  • Adds all-contributors bot to MapReader repo (#297)

Changed

  • New annotation interface in MapReader (#173)

v1.0.6 (2023-11-22)

This release fixes a problem with calculating coordinates of downloaded maps. Releases prior to this will generate incorrect coordinates!

Added

  • Adds images_dir argument when loading annotations, allowing users to specify the path to their patches (#228)
  • Adds pre-commit to the repo to ensure formatting is consistent (#278)
  • Adds option to go straight to inference in the ClassifierContainer class (e.g. with a already fine-tuned model) (#280)
  • Adds metadata_to_save and data_col arguments to the download methods, allowing users to specify additional metadata to save and the column to extract the published date from (#291)

Fixed

  • Warn user if both lower left and upper right corners of the map are missing when downloading maps (#269 and #270)
  • Fixes problem with calculating coordinates of downloaded maps (#276)
  • Fixes deduplication when downloading maps (#285)

v1.0.5 (2023-07-25)

This release marks the end of the Living with Machines project.

Added

  • State dicts are now saved when training/fine-tuning models (#225)

Changed

  • Default delimiter for saving/loading files is now a comma (CSV) (#241)

No changelogs for versions < v1.0.5, see commit messages and PRs instead.

v1.0.4 (2023-06-27)

Pull requests

Commit messages

  • save metadata to csv on each download
  • add error message for broken image files

v1.0.3 (2023-06-21)

Pull requests

Commit messages

  • adds details to dev docs about version numbers

v1.0.2 (2023-06-20)

Pull requests

Commit messages

  • Update Input-guidance.rst w/NLS tile server details
  • takes GH workflows from alto2txt
  • removes poetry as build tool
  • adds missing -m switch
  • checkouts full git history for versioneer.py
  • changes versioneer style
  • fixes production PyPI deployment
  • enables manual triggers

v1.0.1 (2023-06-06)

Pull requests

Commit messages

  • initial commit for workshop notebook
  • fix patchify by meters bug
  • update workshop notebook and fix bug in images.py
  • update notebook and create annotations file
  • update workshop
  • clear outputs and create 'exercise' notebook
  • add show_parent() to workbook
  • add annotations
  • update workshop notebook w/ katies comments
  • update empty notebook

v1.0.0 (2023-06-06)

Pull requests

Commit messages

  • update tests
  • fix coordinates assertion error
  • ensure coordinates are xmin, ymin, xmax, ymax
  • minor updaes to images.py
  • Merge branch '82_align_task_names' into dev_load
  • align to PEP8 style guide
  • Merge branch '110_loader_file_paths' into dev_load
  • Merge branch '116_data_inconsistencies' into dev_load
  • update image_constructor
  • update images_constructor
  • update calc_pixel_width_height
  • update calc_pixel_stats
  • Update images.py remove typo (double underscore infront of _add_shape_id)
  • create parents and patches attributes
  • edits to show/plotting methods
  • move geoinfo method up + split for individual imgs
  • update slicing methods (slicer now obsolete)
  • edits to add_xx and calc_xxx
  • rename/edit add_par to _add_patch_to_parent
  • method to get tree_level from image_id
  • method for verbose printing
  • add patch coordinates
  • auto pull in extra info when constructing images
  • update init + imports
  • run black
  • updated test_loader_filepaths.py
  • make resolving paths a separate static method
  • update tests
  • rename test files
  • rename tests
  • woops sorry - rename tests
  • update tests and add option to ignore_mismatch in add_metadata
  • fix assertions where keys are different
  • rename test
  • update tests for sheet_downloader
  • fix tests
  • Merge branch '110_loader_file_paths' into dev_load
  • update tests
  • add tests
  • fix 'not a PNG file' error
  • Merge branch '116_data_inconsistencies' into dev_load
  • reorder to help with comparision to main
  • update load_xx methods
  • Merge branch '131_geotiff_bug' into dev_load
  • fix rename of update to overwrite
  • fix tests
  • move geo_utils to load subpackage and fix tests
  • fix add metadata tests (checkout file from dev_load)
  • add developers guide to docs
  • Update test_load_loader_add_metadata.py
  • Update contribution guide into tutorials/docs/code
  • update toc trees
  • update toc trees - local only
  • add tqdm and update show
  • Add print info for where patches are saved
  • minor fixes
  • add tqdm to setup.py
  • add beginners info + update input guidance
  • allow metadata as excel file
  • update tests with excel files
  • add error if you try to load non-image files
  • fix tmp_path not defined
  • add beginners info text
  • Update test_load.py
  • add beginners info links
  • save patches as geotiffs
  • save patches as ".tif" only (not ".png.tif")
  • final changes - include CRS as attribute
  • update downloaders
  • fix typo
  • fix tests
  • update saving of coordinates for metadata
  • rename sample tif files
  • update file names in tests
  • rename tests
  • add note about overwriting existing metadata info
  • update reproject_geo_info
  • make images_constructor private
  • rename test_dirs to dirs
  • add specific exception handling
  • add tests for PIL.UnidentifiedImageError
  • add tmp_paths
  • rename proj2convert to target_crs
  • merge test_images and test_images_add_metadata
  • rename download/download2
  • remove download_old
  • update tests and init
  • update extensions
  • update load with advanced usage and to reflect updated code
  • update load title
  • ensure contribution guides included in sidebar
  • add todos
  • minor updates/ typo fixes:
  • Update conf.py
  • fix typo
  • add 'save' argument to convert_images method
  • fix tests
  • Merge branch 'rw_docs' of https://github.com/Living-with-machines/MapReader into rw_docs
  • update docs requirements and conf
  • Update README.md
  • Update Install.rst
  • Update Install.rst
  • Update README.md
  • Update Beginners-info.rst change title
  • make csv default save as tab separated
  • fix xxx_by_string() - allow no keys/keys as string
  • update input guidance
  • update download.rst (user guide)
  • ensure CRS is retained in metadata.csv
  • fix tests
  • raise error if your coordinates are not 4326
  • ensure dictionary keys are same for crs
  • update env name (goodbye mr_py)
  • fix typo
  • Merge branch 'rw_docs' of https://github.com/Living-with-machines/MapReader into rw_docs
  • ensure CRS in patch_df
  • fix sample files
  • update docstrings in dowloader_utils
  • add grid_bb to/from polygon to downloader_utils
  • fix ordering
  • update error message
  • set temp folder in tile_loading + import elsewhere
  • add comment re. temp_folder
  • fix grid_bb to/from polygon (remove self as arguments)
  • update query_by_string tests
  • add tests for crss
  • fix tests
  • fix crs key name
  • add tests for saving dataframes
  • Add tqdm to calc_pixel_stats
  • Create test_sheet_downloader_mock_example.py
  • minor fix to test_sheet_downloader_mock_example.py
  • move load_annotations to learn
  • rename 'kwds' to 'kwargs' to follow convention
  • load_annotations now returns PatchDatasets
  • save to geojson
  • Merge branch '174-save-metadata' into dev_load
  • save patches as geojson
  • update docstrings and ensure PatchDataset methods work
  • create new branch (mock_downloader) for mock downloader and rm file from this branch
  • fix import tests
  • fix tests
  • fix PatchDataset in datasets.py
  • add 'create_datasets' and 'create_dataloaders' methods to 'AnnotationsLoader'
  • set up samplers by default in 'create_dataloader'
  • add todo comment in train docs
  • fix assignment of datasets variable
  • update classifier class to ClassifierContainer
  • add label indices to patch dataset
  • update classifier
  • fix classifier inference
  • allow no label col/label indices in patchdataset
  • add labels map to annotationsloader
  • fix initialise model
  • fix loading preexisting model
  • update doc strings
  • add torchinfo to requirements
  • add tests
  • make shapely version 2.0.0+
  • keep label and pred columns if present when saving to geojson
  • force geopandas to use shapely
  • update tests
  • fix patchify by meters bug
  • rename to classify
  • fix PatchContextDataset
  • fix ClassfierContextContainer
  • Update README.md
  • Update README.md
  • Merge branch 'rw_docs' of https://github.com/Living-with-machines/MapReader into rw_docs
  • update classify name
  • allow use of hugging face (tranformers) models
  • update docs up to training part
  • fix typos
  • Update README.md suggested removal of a line
  • finish updating docs
  • fix docs issues in classifier
  • minor updates to pics
  • add in-out pics
  • fix unique_labels if using append
  • add tests and fix deprecation warning
  • add tests for classifier
  • update pkl files
  • Update test_annotations_loader.py
  • black
  • formatting
  • Merge branch 'dev_train' of https://github.com/Living-with-machines/MapReader into dev_train
  • fix issue with load_patches (ensure patches are added to parents)
  • make work with old annotate
  • Update test_loader_load_patches.py
  • Merge branch 'dev_train' of https://github.com/Living-with-machines/MapReader into dev_train
  • fix some typos and add type hints
  • Update setup.py
  • minor fixes after testing
  • update docs for saving geojson and csv files
  • Merge branch 'rw_docs'
  • fix to show_parent()
  • updates from kmcdono2 comments
  • add doc string and inference guidance
  • fix train_test_split
  • update venv env name from mr_py38 to mapreader
  • update mnist worked example
  • move small_mnist dataset
  • move persistant data (NLS) dataset
  • fix error if using mse loss
  • ensure index has 'image_id' label when saved
  • again fix issue with mse loss
  • update plant pipeline
  • update annotations files
  • remove mnist old notebooks
  • remove coastline example
  • update classification_one_inch
  • update annotation_tasks file
  • update worked examples in docs
  • Update README.md
  • Update conf.py (turn off todos)

v0.3.4 (2023-02-20)

Pull requests

Commit messages

  • version 0.3.3
  • Add CI for pip install
  • change links and citation from arxiv to ACM paper
  • Update README.md
  • initial commit
  • initial commit
  • initial commit
  • adding .readthedocs.yaml
  • add requirements.txt
  • adding README.md
  • update index.rst to include READMEs
  • added READMEs
  • fix requirements.txt
  • setting up docs structure
  • Update Load.rst
  • Update Load.rst
  • adding download section
  • Create how_to_contribute_to_docs.md
  • added basic docs to Download.rst using tutorial as base
  • added basic docs to Load.rst using tutorial as base
  • fix images in Load.rst
  • added basic docs to Annotate.rst using tutorial as base
  • organising docs files into separate directories per 'section'
  • first thoughts
  • updates to train.rst
  • removed process/patchify and testing 'sphinx-disqus' extension
  • added copybutton
  • added copybutton + built html
  • added basic docs to Train.rst using tutorial as base
  • updated Load.rst after going through plant tutorial
  • update Train.rst after going through plant tutorial
  • updates to docs after testing
  • Merge branch 'docs_train' into rw_docs
  • updates to train.rst after testing
  • Update Input-guidance.rst
  • Add issue templates
  • Add versioneer
  • FIx proj and pytorch conda dependancy resolution
  • tweak conda build tests
  • build conda packages for upstream dependancies
  • Successful local build of conda package

v0.3.3 (2022-04-27)

Pull requests

Commit messages

  • Update README.md
  • Move map-related libraries to extras_require
  • Update setup.py
  • Move rasterio to map dependencies
  • Update README.md
  • Update README.md
  • 🧐 adding notebook to branch
  • 🐛 fixing empty notebook
  • Update README.md
  • Update README.md
  • Update README
  • Change version to 0.2.0
  • Move maps notebooks to examples/maps
  • Add MNIST example
  • Add figure for MNIST tutorial
  • Move plant notebooks to non-maps dir
  • Update README
  • Update README
  • Update README
  • Create README.md
  • Update README.md
  • Update README.md
  • Create README.md
  • Update README.md
  • Update README.md
  • Update README.md
  • Update README.md
  • Update README.md
  • BUGFIX: annotation interface when working with images (and not patches)
  • Update README.md
  • Update README.md
  • d test making apr I've proposed removing 'originally' since we want to emphasise the ongoing nature of the library
  • d actual edits to first para I've restored the order to 'maps' -> 'images' so we get a clearer narative as in the current existing repo; and shortened / combined a sentence, as it was repeating 'non-maps' and 'maps', so I used 'any images' instead to make it more intuitive to read. I was also going to add a few sentences giving the nice positive spin about interdisciplinary cross-pollination of image analysis, but not sure where this should go: I don't want to break the flow to the instructions, so perhaps it can go after the bullet points?
  • Merge branch 'restructure-readme-kh' into patch-2
  • minor
  • Change map/non-map to geospatial/non-geospatial
  • Change map/non-map to geospatial/non-geospatial
  • Create README.md
  • update README
  • Merge branch 'restructure-readme-kh' of https://github.com/Living-with-machines/MapReader_public into restructure-readme-kh
  • Move README to geospatial
  • Update README
  • Remove README from classification_one_inch_maps_001
  • Add a new test
  • Add MapReader paper
  • Update README.md
  • black
  • black
  • update CI
  • update CI
  • Add .[maps] to CI
  • Remove as it is empty
  • Version to 0.3.0
  • update README
  • update README
  • update README
  • update README
  • Add windows-latest to CI
  • README: geographic information science
  • Move TOC after gallery
  • Update README.md
  • Add JVC paper
  • Update README
  • Update README
  • Update README
  • Update README
  • Update README.md
  • Set theme jekyll-theme-leap-day
  • Update LICENSE
  • Update README.md
  • Update README.md
  • Update README.md
  • Update README.md
  • Update README.md
  • Update README
  • Update README
  • Update README.md
  • Update README.md
  • Update README
  • Update README
  • Update _config.yml
  • Update README.md
  • Update README.md
  • Update README.md
  • Update README.md
  • Update README.md
  • Set theme jekyll-theme-cayman
  • Set theme jekyll-theme-minimal
  • Update _config.yml
  • Update README.md
  • Update README.md
  • Update README.md
  • change maps to geo in installation
  • review notebooks
  • Add style
  • Update README.md
  • Update README.md
  • Update README.md
  • Update annotations for classification_one_inch_maps_001
  • v0.3.1
  • v0.3.2
  • Update README.md
  • update README
  • add annotations
  • 5 epochs

v0.1.2 (2022-03-03)

Pull requests

Commit messages

  • 📝 adding additional kernel installation instructions.
  • Merge remote-tracking branch 'origin/kasra-hosseini-patch-1' into adjust-poetry-installation-instructions
  • Update README.md
  • add requirements file
  • update poetry files
  • update requirements.txt
  • update requirements.txt
  • update requirements.txt
  • merge with main
  • 🔀 merging branches
  • import torchvision needed for show_sample method
  • Add parhugin v0.0.3 to .toml file
  • 📦 adding requirements.txt file without hashes for binderhub build
  • Merge branch 'dev' of https://github.com/Living-with-machines/MapReader into adjust-poetry-installation-instructions
  • 📦 adding package for binderhub
  • 📦 removing package for binderhub
  • 📌 pinning usage: pyproj [-h] [-v] {sync} ... pyproj version: 3.3.0 [PROJ version: 8.2.0] optional arguments: -h, --help show this help message and exit -v, --verbose Show verbose debugging version information. commands: {sync} version to 3.2.1
  • 📌 downgrading usage: pyproj [-h] [-v] {sync} ... pyproj version: 3.3.0 [PROJ version: 8.2.0] optional arguments: -h, --help show this help message and exit -v, --verbose Show verbose debugging version information. commands: {sync} to 3.2.1
  • 📦 adding to
  • 📦 adding and file to branch
  • Merge branch 'dev' into iss19-model-inference
  • Add max_mean_pixel to the annotation interface
  • Add setup.py, remove .lock and .toml files
  • Update README
  • udpate CI
  • Update setup.py
  • Update setup.py
  • Add quick_start notebook
  • Add files to quickstart notebook

v0.1.1 (2022-01-15)

The first published version of MapReader.