





Finished! Looks like this project is out of data at the moment!
Thank you all for labelling this so fast! If you have appetite for more, please check out our other dataset, which consists of camera shots from vessels:
https://www.zooniverse.org/projects/rvfdevries/camera-ocean-plastic-labeling
General Questions
• Are images classified by multiple people or cross-checked?
Yes, each item must be viewed 5 times (by 5 different users) before it is considered classified/retired. The team can even look at who labeled the data and select users with high classification accuracy.
• What does “retired” or “retirement limit” mean in the Zooniverse statistics page?
Retired means that the classification of the image is complete
Retirement limit is the number of times an image must by classified by different users
• Are there other data sets for us to classify?
Yes! We have lots of images that need classification from a variety of sources (cameras on boats & rivers, drones, etc.)
For now, we are focusing on the images in the Ocean Drones set. Once we get a little further along in the process we will share links to additional data sets that need labeling.
• After the images are labeled, what happens?
Our TOC team gets a list of bounding boxes from each image and the location of each object. The bounding boxes data set is then converted into a format that can be read by an open source machine learning program. This is used to train the neural network to recognize similar objects in new pictures.
• Where were the images that we are classifying taken?
All images in the Oceans Drones set were taken between October and December of 2018 in the “Great Pacific Garbage Patch” (between Hawaii and CA).
• Do I need to create/register for a Zooniverse account in order to help label images?
You can label images without signing in, but it's helpful for us if you sign in. The team is able to see who labeled each image and if there is someone that is a rockstar labeler we can flag that person and prioritize images that they've labeled. But either way, you should be able to classify images just fine.
Specific Labeling Questions & Examples
• Are there images with no debris in them?
Yes. Each original image captured by the drone had debris in it. The images were then split into 4 sections (due to Zooniverse file size restrictions) so it is entirely possible that some of these sections may not have debris in them.
When you see these, do not highlight anything. Simply click next.
• In the Camera on Oceans set I saw a picture with what looked like the side of the boat, do I label that?
No, this is due to the angle of the camera on the boat and should not be labeled as debris.
• What is the difference between fishing gear and ghost net?
Fishing gear are things such as: one string of rope, random buoy, small piece of net, etc.
Ghost nets are one big clump of netting, usually pretty large and looks like a big blob (that’s my super technical description) 😊
• REMEMBER: When in doubt…do not label. If you zoom in a bunch and are still not sure then it is best to leave the image blank so that we don’t confuse the machine learning model.