Neural Twins Talk

TL;DR

This paper introduces Neural Twins Talk, a twin cascaded attention model inspired by human neural pathways, which improves image captioning performance by employing multiple parallel attention channels, demonstrating higher accuracy on COCO dataset.

Contribution

The paper presents a novel twin cascaded attention architecture that enhances image captioning by mimicking human neural pathway expansion, outperforming previous single-channel attention models.

Findings

Higher performance on COCO dataset with multiple attention pathways

Improved image captioning metrics over state-of-the-art models

Model implementation is publicly available

Abstract

Inspired by how the human brain employs more neural pathways when increasing the focus on a subject, we introduce a novel twin cascaded attention model that outperforms a state-of-the-art image captioning model that was originally implemented using one channel of attention for the visual grounding task. Visual grounding ensures the existence of words in the caption sentence that are grounded into a particular region in the input image. After a deep learning model is trained on visual grounding task, the model employs the learned patterns regarding the visual grounding and the order of objects in the caption sentences, when generating captions. We report the results of our experiments in three image captioning tasks on the COCO dataset. The results are reported using standard image captioning metrics to show the improvements achieved by our model over the previous image captioning model. The results gathered from our experiments suggest that employing more parallel attention pathways in a deep neural network leads to higher performance. Our implementation of NTT is publicly available at: https://github.com/zanyarz/NeuralTwinsTalk.