CSV export strategy for Rectangle annotations#122
Conversation
The CSV file can be used with the tooling mentioned here: https://apple.github.io/turicreate/docs/userguide/object_detection/data-preparation.html
|
Hi Kai! In a quick search I could find a "Tensorflow Object Detection CSV format" (e.g. here, basically: I'd still say we can merge your PR for now and I'll change it to the Tensorflow format in a separate PR later on. |
|
Can you run the |
|
Done that. Wow, I didn't know that the CSV format was SO special. That was just a brief excursion into this topic for me. |
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #122 +/- ##
============================================
- Coverage 86.46% 86.41% -0.06%
- Complexity 1480 1486 +6
============================================
Files 95 96 +1
Lines 5933 5970 +37
Branches 534 540 +6
============================================
+ Hits 5130 5159 +29
- Misses 527 535 +8
Partials 276 276 ☔ View full report in Codecov by Sentry. |
While I was following this guide I needed BoundingBoxEditor to export the annotations to a CSV file. As the format is pretty specific, I'm not sure if it may be a good addition for this project. But maybe it can be extended to be more flexible in the future. The exporter can only export Rectangle annotations.
What do you think, can this be merged?
I introduced two new dependencies: OpenCSV for writing the CSV file and Jimfs for the unit tests.