Animals 151

We gathered a small dataset of images to test out how well different models can classify them. Each category in the dataset has at most 60 images, and some have only 30. To keep things manageable, We resized all the images to 224×224 pixels.

What we want is to develop a highly accurate model using a small amount of data.

You can take a look at some example notebooks for ideas. We’ve also provided a translation file called translation.json, which converts scientific names of animals into common names. This will help you understand the names of the animals you’re identifying in simpler language.

We usually add new data to this dataset about once a month, but there might be times when We don’t update it for a while. This is just a hobby for us, not something We do full-time.