← Back to Publications

Conditional GANs for Image Captioning with Sentiments

T. Karayil, A. Irfan, F. Raue, J. Hees, A. Dengel

International Conference on Artificial Neural Networks (ICANN)||DOI: 10.1007/978-3-030-30490-4_25
GANsImage CaptioningSentiment AnalysisDeep Learning

Abstract

We explore the use of conditional Generative Adversarial Networks for generating image captions that incorporate sentiment, combining visual understanding with affective language generation.

Overview

Image captioning traditionally focuses on generating factually accurate descriptions. In this work, we extend the task to include sentiment, generating captions that not only describe what is in an image but also convey a specific emotional tone.

Method

We propose a conditional GAN-based approach that conditions the caption generation process on both visual features and desired sentiment. This allows for controllable generation of affective image captions.