Neural Image Captioning: Deep Learning for Visual Description

This coursework explores Neural Image Captioning (NIC), covering its importance, challenges, and advancements over traditional methods. The NIC model combines CNN and RNN architecture for image-text alignment, utilises beam search for coherent captions, and is validated on major datasets. Experimental evaluations show significant improvements in BLEU scores and human judgments. Comparisons highlight NIC's superiority over previous rule-based systems, inspired by neural machine translation....