search

microsoft / VoTT

Visual Object Tagging Tool: An electron app for building end to end Object Detection Models from Images and Videos.
MIT License
4.3k stars 837 forks source link

Format Converter on annotations #151

Closed kyoro1 closed 6 years ago

kyoro1 commented 6 years ago

As we find in README.md in the top folder, we have some formats:

Exporting tags and assets to CNTK , Tensorflow (PascalVOC) or YOLO format for training an object detection model.

Then, if we have already completed adding annotations for Faster R-CNN(CNTK), can we have a kind of format-convert to Yolo? (Or other directions like Yolo to CNTK...) Perhaps, is it required to generate by ourselves?

aribornstein commented 6 years ago

Did you generate your cntk annotations with Vott? If so should be able to just reload the Annotations into Vott and just export them as yolo.

nolancardozo13 commented 6 years ago

Any update on this? I have CNTK annotations generated by VOTT and would like to use these to convert to YOLO or Pascal VOC. I do not think VOTT has this functionality at present.

aribornstein commented 6 years ago

Yes this is something i have added to the backlog what i recommend is that you share the .json file generated by vott in the same directory as your image and video assets that should save you from having to reproduce all the tags.