We launched two projects – ‘Etch A Cell – Fat Checker’ in February 2021 and ‘Etch A Cell – Fat Checker Round 2’ in March 2022 to ask Zooniverse volunteers to help us examine fat droplets in microscopy images. Over the course of these two projects, more than 2000 Zooniverse volunteers contributed over 75 thousand annotations!  Thank you!

Since the successful completion of these projects, we have been working hard to analyze the data contributed by Zooniverse volunteers. Your efforts have enabled us to train various machine learning models for fat droplet identification. In the images shown above you can see the results of one of these machine learning models – in the top row we show a single 2D image from our original data set alongside the collective volunteer  annotation of the lipid droplet, we then show three images from different machine learning models predicting the location of the lipid droplets (in green, yellow and blue). You can read more about this work here.

In the future we aim to make projects similar to this one more efficient. To do this, we have added new  functionality to the drawing tool – volunteers will be able to edit, undo, redo, auto close, or delete a drawn freehand shape. Go check this out at Elephant ID – our first project to use this new tool! Next up – we are working towards you being able to correct the machines. We’ve got several projects (hopefully) coming out soon which show machine-based outlines as starting points for users to suggest corrections. In this way we are making the best partnership between humans and machines for science.