Camera Traps (or Wild Cams) enable the automatic collection of large quantities of image data. Biologists all over the world use camera traps to monitor biodiversity and population density of animal species.As biologists try to expand the scope of these models from specific regions where they have collected training data to nearby areas they are faced with an interesting probem: how do you classify a species in a new region that you may not have seen in previous training data?

Data is primarily provided by Erin Boydston (USGS), Justin Brown (NPS), iNaturalist, and the Idaho Department of Fish and Game (IDFG). The training data and test data are from different regions, namely The American Southwest and the American Northwest.

The test set contains 153,730 images from 100 locations in Idaho. The location id (location) is given for all images.

The set of training classes is: 'bobcat', 'opossum', 'coyote', 'raccoon', 'dog', 'cat', 'squirrel', 'rabbit', 'skunk', 'rodent', 'deer', 'fox', 'mountain_lion', 'empty'

The set of test classes is: 'bobcat', 'opossum', 'coyote', 'raccoon', 'dog', 'cat', 'squirrel', 'rabbit', 'skunk', 'rodent', 'deer', 'fox', 'mountain_lion', 'moose', 'small_mammal', 'elk', 'pronghorn', 'bighorn_sheep', 'black_bear', 'wolf', 'bison', 'mountain_goat', 'empty'

There are 23 classes for which need to make a classification or in other words "multiclass classification problem". Using external data is allowed.

You can find the iWildCam 2019 Competition here or here.
There is a github page for the competition here.