Bulk Animal Classification In the Wild...

This is the video demo on doing bulk animal classification using deep learning to help research to classify the animals in the images. The researchers will upload folder where they have bulk images (gigabytes of images) and the model will run prediction algorithm by looping over every image in the folder and store the images with animals in a separate folder and image with no animals in a separate folder...

[@Neplzguy]

Great job mate …

[@MrCorpsy6]

All those miss classifications come from what we call "shortcut learning". For instance, at 4:14 it's clear that your model thought it was a picture with an animal because it's almost the same picture than the one with the monkey. It does not know how to detect an animal, but instead it noticed strong correlations between the labels and the background of the pictures. Shortcut learning is why most of those models fail in practice.