Mea dev

CHANGELOG.md

@@ -1,11 +1,19 @@
 # Changelog
 
+# Version 2.7b0+mea (2021-06-27)
+## Fix
+* Update `AnnotationGroup` to expect labelId to be a cuid instead of uuid.
+* Update `datarow_miou` to support masks with multiple classes in them.
+
+
+# Version 2.7.0 (2021-06-27)
+## Added
+* Added `dataset.export_data_rows()` which returns all `DataRows` for a `Dataset`.
 
 # Version 2.6b2+mea (2021-06-16)
 ## Added
 * `ModelRun.annotation_groups()` to fetch data rows and label information for a model run
 
-
 # Version 2.6.0 (2021-06-11)
 ## Fix
 * Upated `create_mask_ndjson` helper function in `image_mal.ipynb` to use the color arg

docs/source/conf.py

@@ -21,7 +21,7 @@
 copyright = '2021, Labelbox'
 author = 'Labelbox'
 
-release = '2.5.6'
+release = '2.6.0'
 
 # -- General configuration ---------------------------------------------------
 

docs/source/index.rst

@@ -14,6 +14,13 @@ Client
    :exclude-members: upload_data, upload_file
    :show-inheritance:
 
+AssetAttachment
+--------------------------------------
+
+.. automodule:: labelbox.schema.asset_attachment
+   :members:
+   :show-inheritance:
+
 AssetMetadata
 --------------------------------------
 

examples/integrations/tlt/README.md

@@ -0,0 +1,20 @@
+# NVIDIA + Labelbox
+
+##### Turn any Labelbox bounding box project into a deployed service by following these tutorials
+
+--------
+
+
+#### labelbox_upload.ipynb
+* Download images and prelabels
+* Setup a labelbox project
+* Upload prelabels to labelbox using MAL
+* Clean up the data in labelbox
+
+#### detectnet_v2_bounding_box.ipynb
+* Plug in training data from previous step (or bring your own labelbox project)
+* Train a model using TLT. Compare with a non-pretrained model
+* Prune the model for more efficient deployment
+* Convert the model to a TRT engine
+* Deploy the model using Triton Inference Server 
+