A Comparative Study on Optimizers for Automatic Image Captioning

In the field of artificial intelligence, computer vision and natural language processing are used to automatically generate an image's contents. The regenerative neuronal model is developed and is dependent on machine translation and computer vision. Using this technique, natural phrases are produced that finally explain the image. This architecture also includes recurrent neural networks (RNN) and convolutional neural networks (CNN). The RNN is used to create phrases, whereas the CNN is used to extract characteristics from images. The model has been taught to produce captions that, when given an input image, almost exactly describe the image. The outcome of these algorithms is determined by several factors, including feature extraction, caption generation, and optimizer selection. Our goal is to conduct a comparative analysis of several optimizers to determine the optimizer that achieves highest accuracy for a deep learning model. The deep learning model is subsequently trained with various optimizers on the Flicker dataset. The accuracy of the results of the model using optimizers are achieved as follows: RMSprop optimizer has a 92% accuracy, SGD has a 12% accuracy, Adam optimizer has 53% accuracy, and Adadelta has a 12% per cent.