ImProvShow: Multimodal Fusion for Image Provenance Summarization


Alexander Black (University of Surrey), Jing Shi (Adobe Research), Yifei Fan (Adobe Research), John Collomosse (Adobe Research)
The 35th British Machine Vision Conference
PDFPosterVideo (Right click to download)Supplementary

Abstract

We present ImProvShow; a novel approach to summarizing the multi-stage edit history (or `provenance') of an image. ImProvShow fuses visual and textual cues to succinctly summarize multiple manipulations applied to an image in a sequence; a novel extension of the classical image difference captioning (IDC) problem. ImProvShow takes as input several intermediate thumbnails of the image editing sequence, as well as any coarse human or machine-generated annotations of the individual manipulations at each stage, if available. We demonstrate that the presence of intermediate images and/or auxiliary textual information improves the model's edit captioning performance. To train ImProvShow, we introduce METS (Multiple Edits and Textual Summaries) - a new open dataset of image editing sequences, with textual machine annotations of each editorial step and human edit summarization captions after the 5th, 10th and 15th manipulation.

Citation

@inproceedings{Black_2025_BMVC,
author    = {Alexander Black and Jing Shi and Yifei Fan and John Collomosse},
title     = {ImProvShow: Multimodal Fusion for Image Provenance Summarization},
booktitle = {36th British Machine Vision Conference 2025, {BMVC} 2025, Sheffield, UK, November 24-27, 2025},
publisher = {BMVA},
year      = {2025},
url       = {https://bmva-archive.org.uk/bmvc/2025/assets/papers/Paper_1081/paper.pdf}
}


Copyright © 2025 The British Machine Vision Association and Society for Pattern Recognition
The British Machine Vision Conference is organised by The British Machine Vision Association and Society for Pattern Recognition. The Association is a Company limited by guarantee, No.2543446, and a non-profit-making body, registered in England and Wales as Charity No.1002307 (Registered Office: Dept. of Computer Science, Durham University, South Road, Durham, DH1 3LE, UK).

Imprint | Data Protection