The problem to solve is to classify 6 species of fish in real world images with varying backgrounds
It applies Haar Cascade object detection algorithm and 16-layer Convolutional Neural Network.
The CNN has architecture similar to that of VGG-16, with different hyper-parameters.
It contains work by Ming Li for Kaggle Challenge.