Annotating Images with Microsoft vott

Chances are you'd have to collect your own data for a custom object detection or object segmentation model at some point. And unlike image classification tasks, you cannot simply feed the model with unlabelled images.

In 10 minutes I walk through the options you have for annotations, how to label with microsoft vott and the export options it offers. There isn't a one-software-fits-all when it comes to data annotation so your needs may differ a little bit.

Consider the other applications I mentioned if you're annotating videos or need an AI assisted annotation tool for large (e.g. 50,000+ images) datasets.
---------------------
Chapters
00:00 intro
00:38 data annotation options
01:04 why I use ms-vott
01:27 installing ms-vott
02:10 launch application
03:42 drawing bounding boxes for object detection
04:32 polygonal annotations for segmentation
05:48 best practices
07:11 deleting annotations
07:30 export
09:19 wrapping up
-------------------
LINKS
best practices for bounding boxes:
general best practices for data annotation:
--------------------
About me:
--------------------
#computervision #datasets #objectdetection #imagesegmentation

[yuvraj]

Thanks Tomi for this accurate tutorial !

[kezstain]

Awesome reminder. They don't document it will but you can type the number next to the tag after selecting a region 4:08