Image Classification Using TensorFLow
I did a project in Machine Learning for an image classification task in which we designed a model using two layer neural networks to train the dataset of images containing either cats or either dogs which was downloaded from kaggle's Kernels Edition.
I used python and used TensorFlow to build the network of my model.
In this project, I initially trained my model using the original rgb dataset from kaggle and then I converted the rgb dataset into images containing only one channel for H value once and again did the same only V value. At the end I compared the training accuracy among the three cases stated.
I found according to my model which is of only two layered neural network, training accuracy came almost the same in the range of 50-60%.
Accuracy might have increased or varied if I would have increased the no.of layers in my model or might have used a Convolution Neural Network(CNN).