Annotations for images

[arossbach (André Roßbach)]

What are the options for annotating images with Prodigy (bounding boxes, segmentation)?
Do single points with a certain meaning also work, e.g. Facial Landmark?

Best regards
André.

[ines (Ines Montani)]

Hi! You can find more details on the image annotation capabilities here:

Computer Vision · Prodigy · An annotation tool for AI, Machine Learning &...

To see the interface in action, you can also try the live demos:

• Manual image annotation
• Binary object detection (e.g. with a model in the loop)
• Binary image segmentation (e.g. with a model in the loop)
• Multiple choice with images

This thread has some examples and ideas for how write a custom recipe to plug in a computer vision model and annotate with a model in the loop:

Using Prodigy to train a new Computer Vision object detection model

Hi! I’ve begun taking a look at prodigy’s computer vision abilities. I see that it works with Lightnet which has a set of ~80 objects it can detect. It seems like you can update prodigy to work with any of those existing objects/categories, but how would I train the model to recognize a new object entirely? For example, say I want to build an object detection model for a stapler. Would I try to adapt one of the existing categories to this new definition or would I try to add a new category enti…

@arossbach

Do single points with a certain meaning also work, e.g. Facial Landmark?

Manual image annotation currently only supports rectangles (i.e. for bounding boxes) and polygons. Adding single points could be a nice addition, though!

The manual image interface is still quite new and not 100% feature complete yet. In general, Prodigy focuses very much on the iterative development workflow, scripting custom annotation flows and connecting the annotation process with machine learning models. If you only need to draw points or shapes onto images, you might find that a much simpler tool is actually a better fit for what you're trying to do.