Deep Learning wildlife classifier - try it yourself

In this image our model correctly recognizes the Grant's gazelle, a Thomson's gazelle, and the zebras in the back

In this repo we share the model we have trained and used to obtain 5th place in the Hakuna Ma-data competition.

Try it out yourself!

To play with it you can:

• clone the repo, and then use the classify_images notebook
• open the classify_images_on_colab notebook on Google Colab (you can use this link) and enjoy the model without the need to setup the environment (try uploading your own data!)

We have provided a few images from the Snapshot Serengeti project (on which the model was trained), and a sample of fun examples. Let us know if you find something cool or funny!

Note:

The classes our model is trained to recognize are:

aardvark, aardwolf, baboon, bat, batearedfox, buffalo, bushbuck, caracal, cattle, cheetah, civet, dikdik, duiker, eland, elephant, empty, gazellegrants, gazellethomsons, genet, giraffe, guineafowl, hare, hartebeest, hippopotamus, honeybadger, hyenaspotted, hyenastriped, impala, insectspider, jackal, koribustard,leopard, lionfemale, lionmale, mongoose, monkeyvervet, ostrich, otherbird, porcupine, reedbuck, reptiles, rhinoceros, rodents, secretarybird, serval, steenbok, topi,vulture, warthog, waterbuck, wildcat, wildebeest, zebra, zorilla