I have created a large data set of images (800 odd) which I’m annotating by hand. I noticed that the Roboflow help info indicates there is an Auto Label option in the Upload area but it’s not being displayed, I only seem to have the manual annotation option. Can anybody tell me how to get access to the auto annotation function?
Hi there - what project type are you using?
You are likely using a project type that doesn’t support Auto Label (like Multimodal).
If you let me know where in the UI you’re seeing that helper text let me know and I can clarify it.
Hi Jacob,
I’m relatively new at this, what do you mean by project type? My project is called Bucket Tracking and using Key Point Detection - https://app.roboflow.com/tracking-test-yqcy2/bucket-tracking/
I seem to recall there was a menu available when I first started messing around with Roboflow, but steep learning curve and blur… I just did a search on Auto Label and there’s a whole pile of useful info but my set up only allows ‘Manual Annotation’.
I may have sorted my own problem out…
This began as a ‘what if’ project, not expecting it to work…BUT it works so well, it’s now a hurry up and produce a decent sized database. About 2000 images will do for each of the 4 classes, but that’s a lot of work… Anyway, I figured out how to use my already trained model to run through a mp4 video and pull out and identify every nth frame. It then creates a file name and loads the image into the correct directory based on the class it identified. This speeds the annotation process up greatly as each directory contains images that are similar and the annotations just need some tweaking after using the ‘Repeat Previous’ button…
Then in a ‘light bulb’ moment I realised it produces the box and key point coordinates with each new inference, I need to figure out how to create the correct formatted file to import the new file directly into the import section somewhere. The new images will already be classified and annotated. All I need to do is a quick check for accuracy and add them to the data set??? If that makes any sense???
Question about annotating?
In some of my images the Key Points are partially obscured, and some slightly off the image all together. Generally I mark them as ‘Obscured’ if they’re in the image. Should I be deleting the keypoint if it’s not on the image??
Thanks!
Brent Simpson
BTW: Roboflow is awesome…
1 Like
Hi @automateme2 - thanks for all the context and I’m glad you’re surprising yourself with good results 
Project type is the same as task type, this can include Object Detection, Instance Segmentation, Classification, etc. In your case, it’s Keypoint Detection.
Unfortunately, Auto Label is not supported currently with Keypoint Detection (I’ve flagged this internally, we may be able to push out a quick fix).
For your labeling flow, you might have a much easier time using Workflows – we literally built it for cases like this. You can run a model and selectively upload images directly to Roboflow with the correct annotation format. You’re correct that you should be manually reviewing images before adding them to your dataset.
On the question of obscured keypoints, the rule of thumb is you should label images exactly how you would like your model to respond to the given input. If you want it to not give a keypoint, mark it obscured.
@automateme2 - I talked with the team, and unfortunately adding Auto Label for keypoint is not a quick fix and we’re not able to prioritize it in the next ~month.
If others are running into this and want to see the feature, let us know! Will make it easier to prioritize.
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.