Texture-based latent space disentanglement for enhancement of a training dataset for ANN-based classification of fruit and vegetables

The capability of Convolutional Neural Networks (CNNs) for sparse representation has significant application to complex tasks like Representation Learning (RL). However, labelled datasets sufficient size learning this are not easily obtainable. unsupervised Variational Autoencoders (VAEs) and Generative Adversarial (GANs) provide a promising solution issue through their capacity learn representations novel data samples classification tasks. In research, texture-based latent space disentanglement technique is proposed enhance samples. A comparison performed among different VAEs GANs with the approach synthesis new Two VAE architectures considered, single layer dense convolution based VAE, compare effectiveness representations. selected on distance metric disjoint distribution divergence estimation been shown improvement disentangling process by conditioning random noise synthesising texture rich images fruit vegetables.