Evaluation of neural networks and feed forward neural network models on to content-based image retrieval

Abstract

The advanced technological developments in the machine learning models are being to develop new methodologies for content based image retrieval (CBIR). Since the ML models has the capability of learning global visual features for any given query enables them a better solutions for the models deal with massive amount of different image dataset. At the same time, the application of ML models like neural networks (NN) has some difficulties like the search goal has to be fixed or the computation complexity become too expensive for an online setting. In this study, a performance evaluation is carried out between NN and feed forward neural network (FNN) for CBIR. A set of benchmark images is employed to study the performance of the two ML models interms of different measures. The attained results exhibit that the FNN model is found to be better than the NN on all applied test images.