top of page

This image labelling challenge is now closed

Annotate images of fish collected from Billabongs in Kakadu National Park, Australia

Scientists monitoring fish community health need your help to label 500,000+ images of fish found in billabongs! Labelled images will be used to train artificial intelligence for automating the identification and counting of fish!  



How to get started!

Images are labelled in Microsoft's Visual Object Tagging Tool (VoTT) using the web browser version at the bottom of this page.


To start your labelling journey follow these simple steps from the VoTT home page:

Connect VoTT to your images:

1. Access VoTT at the bottom of this page and navigate to the connections window.

2. From the connections window press the + symbol to add a new connection.

3. Click on the provider dropdown box and select Azure Blob storage.

4. Fill in the following default values:

  • Display Name: Your name to identify your labels

  • Account Name: Provided in email request to participate

  • Container Name: Provided in email request to participate

  • SAS: This is the unique code provided in your email request to participate

5. SAVE your connection.

Add security token:

1. Click on the gear symbol in the bottom left of the VoTT page

2. Click add security token

3. Copy and paste the following from your email:

  • Name: citizenscientist Token

  • Key: Provided in email request to participate

Open a cloud project:

1. Click on the house icon to return to the home page and open a cloud project.

2. Click on the citizenscientist cloud connection, then OK.

3. Click on citizenscientist.vott file, then OK



Your images will appear, you can now begin drawing polygons around fish in the images! 

Annotate Images:

1. With the current image open, click on the polygon tool from the toolbar at the top of the screen (third from the left)

2. Position the mouse pointer on the outline of the fish in your image and click the left mouse button to start adding vertices. When you've added points around the fish double click the left mouse button to finish and close the annotation.

BONUS! If you'd like to put your fish ecology skills to the test, try placing a species name to your tags! Watch the video below  to see how we annotate and label.


Labelling Guidelines


  • Only label fish that you can confidently see the outline of when staring at the image. If you're unsure,  squint at the image and see if you can still see the fish clearly, if not, do not label.

  • Try to use the least number of polygon points as possible to capture the entire fish, besides giving you a sore hand, too many points is unnecessary and time consuming. 

  • If the fish is obstructed by either another fish or piece of debris, avoid creating two separate polygons for the fish. Rather, annotate the head and remainder of body up until it is obscured. 

  • Large schools of small fish will be visible in the background, however not as clearly as those up close. Avoid labelling small specks, this will introduce noise into the dataset.

Freshwater Fishes of Kakadu

These are some of the species you may encounter in your images! If you think you know what fish you have put a label on it! 

Toxotes chatareus (Archer Fish)
Strongylura kreffti (Freshwater Longtom)
Scleropages jardinii (Saratoga)
Pseudomugil gertrudae (Spotted Blue-eye)
Pseudomugil tenellus (Delicate Blu-eyemale)
Porochilus rendahli (Randahl's Catfish)
Pingalla midgleyi (Midgleyi's Grunter)
Oxyeleotris lineolata (Sleepy Cod)
Neosilurus ater (Black Catfish)
Nematolosa erebi (Bony Bream)
Melanotaenia splendida inornata (Chequered Rainbowfish)
Mogurnda mogurnda (Northern Purplespotted Gudgeon)
Megalops cyprinoides (Oxeye Heron)
Leiopotherapon unicolor (Spangled Grunter)
Hephaestus fuliginosus (Sooty Grunter)
Lates calcarifer (Barramundi)
Glossamia aprion (Mouth Almighty)
Denariusa bandata (Penny Fish)
Amniataba percoides (Banded Grunter)
Craterocephalus stercusmuscarum (Fly-specked Hardyhead)
Neoarius spp.
Ambassis agrammus (Sailfin Glassfish)
Classification Tree.png

VoTT! Your Labelling Tool

You may need to zoom out on your screen for full view.

Supported web browsers include Google Chrome and MS Edge (Latest Version)

bottom of page