UmojaHack #1: SAEON - Identifying marine invertebrates (ADVANCED) by UmojaHack Africa
$1,000 USD
Deep sea invertebrates image classification challenge
279 data scientists enrolled, 81 on the leaderboard
ConservationResearchComputer VisionStructured
Africa
21 March
7.5 hours

The objective of this challenge is to correctly classify images of deep sea invertebrates.

The images are of various species of marine invertebrates found off of the coast of South Africa. Further details of the species can be found in the linked field guide.

There are 3111 images in the train set, 1423 images in the test set and 137 different species. The test set has the same class distribution as the training set.

Here are some examples of the species and their scientific names.

In order to make access to the data easier for all participants, we have provided download links. We recommend you download the data before the challenge. The data is password protected, and we will share the password to all universities as well as on the livestream when the competition opens.

Folder codes will be shared on the day at 09:00 GMT on the WhatsApp groups

Files available for download:

  • Train.zip - his zip file contains 137 folders: one for each specie. These are the categories you will use to train your model.
  • Test.zip - contains 1434 images that you will implement your model on to classify each species.
  • SampleSubmission.csv - is an example of what your submission file should look like. The order of the rows does not matter, but the names of the IDs must be correct.