This is actually just a product of the system not correctly detecting the body. We manually re-ran the image through detection and it still did not pick up on the body.
We are currently finishing up alpha testing on a manual annotation tool. Once we finish, it will become available for general use as a beta feature. I need to finish up documenting the feature capabilities and limitations, and then we will push it live. Goal is to have it ready by the end of the week. We will reply back here once it’s available on ACW (tracking with WB-671).
We’ve successfully moved the manual annotation tool to to beta. Documentation for it can be found here.
I’m going to cover some very important points here as well.
To see the manual annotation option, you will need to perform a hard refresh on your browser (Ctrl+F5). The option is found on the encounter menu page.
At this the manual annotation tool only supports body annotation, not part annotation.
When creating an annotation, set both dropdowns before creating the annotation; if you create the annotation first, the annotation can shift on the asset, and then that messes up your data and makes a match less likely and makes it unhelpful training data.
Let me know if you have any questions!