Skip to content

Releases: v7labs/darwin-py

v0.7.20

30 Aug 14:29
efd324e

Choose a tag to compare

Fixed issue where only default team could be used when importing into V2 dataset.

v0.7.19 Released

30 Aug 11:11
2707abb

Choose a tag to compare

Added support for importing into darwin v2 datasets

v0.7.18

29 Aug 09:07
1e353a8

Choose a tag to compare

  • Add support for conversion of Darwin JSON annotations to YOLO format

v0.7.17

26 Aug 12:19
d199a0f

Choose a tag to compare

  • Only installs dataclasses if python version 3.6 or older. (thanks @campbell-remedy)
  • Fixes issue with dataset creation in v2

V0.7.16

10 Aug 16:43
cc96939

Choose a tag to compare

Improves error message when API call fails.

v0.7.15

02 Aug 10:48

Choose a tag to compare

Adds support for /api/v2, the sdk api at the surface level should be unchanged, but then underlying structure has gone through some large changes.

RemoteDataset is now an abstract class with implementations for both v1 and v2 datasets, clients picks the right implementation depending on payload.

Client also has a new .api_v2 field, that contains all backend calls to /api/v2, I would like to do the same for v1 but it's possible that past users are depending on tho
se calls existing inside client which makes this a bit tricky.

v0.7.14

26 Apr 13:28
17f41b9

Choose a tag to compare

  • Fixed coco imports not working (#399)
  • Completed SDK documentation (#400, #402, #403)
  • Platform now allows repeated class names, as long the instances have different types (#401)

v0.7.13

13 Apr 13:07
1d279cf

Choose a tag to compare

  • Deleting non-existent files from a dataset now shows a warning message. Updated delete-files success message to better reflect the operation. (#393)
  • Fixed exports not filtering by class ids. Updated docs. (#394)
  • Fixed val-percentage and test-percentage being optional when they were required. Updated -h menu message. Now when using dataset split you have to use: darwin dataset split dataset-name -v X -t Y (#388)
  • Added option to have Rich output when using dataset report. Ex.darwin dataset report dataset-name -r (#387)
  • Listing files now prints a rich table (#385)
  • Several documentation improvements and API trimming (#373, #382, #384, #380, #374, #375, #376, #377, #378, #379, #397)
  • Fixed typos and incorrect menu -h descriptions (#386, #390)
  • Added export flag for including authorship. Ex. darwin dataset export dataset-name export-name --include-authorship (#389)
  • Reverted Release download requiring a client to download a zip (#372)
  • Added warning for class collisions when importing (#368)

v0.7.12

10 Mar 18:11
639226a

Choose a tag to compare

Revert changes to Release's download_zip method, so it does not assume a config file exists for Darwin.

The users should not see any MissingConfig errors being raised when pulling a dataset via Darwin-Py API.

v0.7.11

01 Mar 12:00
7a0556f

Choose a tag to compare

  • Hotfix: supporting python 3.6 again
  • Added docs and trimmed API surface of darwin/exporter/formats/pascalvoc.py
  • Added docs to darwin/exporter/formats/instance_mask.py
  • Added docs to darwin/exporter/formats/semantic_mask.py