Self-supervised Cross-view Representation Reconstruction for Change Captioning
Abstract
Change captioning aims to describe the difference between a pair of similar images. Its key challenge is how to learn a stable difference representation under pseudo changes caused by viewpoint change. In this paper, we address this by proposing a self-supervised cross-view representation reconstruction (SCORER) network. Concretely, we first design a multi-head token-wise matching to model relationships between cross-view features from similar/dissimilar images. Then, by maximizing cross-view contrastive alignment of two similar images, SCORER learns two view-invariant image representations in a self-supervised way. Based on these, we reconstruct the representations of unchanged objects by cross-attention, thus learning a stable difference representation for caption generation. Further, we devise a cross-modal backward reasoning to improve the quality of caption. This module reversely models a ``hallucination'' representation with the caption and ``before'' representation. By pushing it closer to the ``after'' representation, we enforce the caption to be informative about the difference in a self-supervised manner. Extensive experiments show our method achieves the state-of-the-art results on four datasets. The code is available at https://github.com/tuyunbin/SCORER.
Cite
@article{arxiv.2309.16283,
title = {Self-supervised Cross-view Representation Reconstruction for Change Captioning},
author = {Yunbin Tu and Liang Li and Li Su and Zheng-Jun Zha and Chenggang Yan and Qingming Huang},
journal= {arXiv preprint arXiv:2309.16283},
year = {2023}
}
Comments
Accepted by ICCV 2023