1.1.10

This release adds a new CLI tool, so the user can download and upload from the command line

Video inference bug-fix release

This release fixes some bugs in the previous release re: video inference.

1.1.8

This release introduces video inference bindings for the Roboflow Python package. These bindings allow users to use the hosted Roboflow Video Inference solution.

1.1.7

Updated python version requirements to >=3.8 from 3.6 in setup.py and Readme. Python 3.8 is now the earliest version Roboflow supports because it’s the earliest version supported by our dependencies. Here’s roboflow/supervision#180 from when supervision dropped support for 3.7 a few months ago.

1.1.6

What's Changed

• Add input continue warning for wrong ultralytics version in model deploy by @SolomonLake in #185

Full Changelog: v1.1.15...v1.1.6

v1.1.4

bugfix: project.upload() was using a wrong parameter name for is_prediction

v1.1.3

Upload of large datasets with workspace.upload_dataset() is more reliable.

• Doesn't break upload in the middle even if it gets unexpected responses from the backend
• Better separation of output to stdout and stderr (stderr should only have the list of problematic files)
• Ability to resume uploads to the same project (if the workspace has a project with a matching name)

v1.1.2

Fixed logging for the version deploy method.

v1.1.1

In this release we add API support for search.

v1.1.0

In this release we add a feature to upload VOC and YOLO formatted datasets to Roboflow

@SkalskiP will be excited about the supervisioninclusion` which made the YOLO uploads possible.

You can turn up the parallelism to speed up dataset upload speed, but there must be some limit to that.