sibyjackgrove/CNN_for_bird_classification

CNN to classify images of birds into Cardinal and Indigo

CNN_for_classifying_bird_images

Convolutional Neural Network to classify images of birds into Cardinal and Indigo. Only TensorFlow operations are used to read image files, and build CNN.

Dependencies

TensorFlow 1.0, Numpy

Operations

tf.layers, tf.image, tf.gfile, tf.losses

CNN Architecture

2 Convolutional layers with ReLU activation, 2 Max Pooling Layers, 1 Fully Connected Layer with Softmax Activation

Training

Training data: 72 images, Validation data: 24 images, Testing data: 4 images

Classification - 3 Classes

0: Cardinal; 1: Indigo; 2: None of these

Dataset

Source: CUB-200 available at http://www.vision.caltech.edu/visipedia/CUB-200.html