This Deep Learning project is a Computer Vision project. The objective is, from an image of a bird, to find its species, among the 260 species of birds in the database.
The chosen data set includes 40 000 images of birds of 260 different species.
To accurately predict the species of a given bird, we have :
- built different types of neural networks: simple perceptron, multi layer perceptron, convolution network, transfer learning: ResNet18, GoogleNet, AlexNet, VGG16, ImageNet.
- used other approaches: feature extraction, then application of Sector Vector Machine (SVM), Random Forest, K-Nearest Neighbors and SGD
The best accuracy obtained is 94% with the AlexNet network.