Amharic Language Image Captions Generation Using Hybridized Attention-Based Deep Neural Networks

This study aims to develop a hybridized deep learning model for generating semantically meaningful image captions in Amharic Language. Image captioning is task that combines both computer vision and natural language processing (NLP) domains. However, existing studies the English primarily focus on visual features generate captions, resulting gap between textual inadequate semantic representation. To address this challenge, proposes attention-based neural network (DNN) model. The consists of an Inception-v3 convolutional (CNN) encoder extract features, attention mechanism capture significant bidirectional gated recurrent unit (Bi-GRU) with decoder captions. was trained Flickr8k BNATURE datasets which were translated into Language help Google Translator experts. evaluation showed improvement its performance, 1G-BLEU score 60.6, 2G-BLEU 50.1, 3G-BLEU 43.7, 4G-BLEU 38.8. Generally, highlights effectiveness hybrid approach better meaning.