acmmm

name	acmmm
description	Writing and formatting guidance for ACM MM (ACM International Conference on Multimedia) papers. Use when drafting, formatting, or submitting to ACM MM 2026. Covers ACM acmart sigconf format with double-blind review mode, ACM MM page limits (8 pages), multimedia-specific conventions (multi-modal content, figures, supplementary materials), and ACM camera-ready preparation.

ACM MM 2026 Conference Writing Skill

CCF-A | Multimedia | Publisher: ACM Conference: https://www.acmmm.org/2026 Template: ACM-MM/acmart.cls (ACM acmart, sigconf format)

Document Setup

Preamble Structure

% === SUBMISSION MODE (anonymous, double-blind) ===
\documentclass[sigconf, review, anonymous]{acmart}
\settopmatter{printfolios=true}

% === CAMERA-READY MODE ===
\documentclass[sigconf]{acmart}
\acmConference[ACM MM 2026]{MM '26: 2026 ACM International Conference on Multimedia}
               {October 13--17, 2026}{Singapore}
\acmISBN{978-X-XXXX-XXXX-X/XX/XX}
\acmDOI{10.5555/XXXXXXX.XXXXXXX}

Required Packages

\usepackage[round]{natbib}     % Author-year citations
\usepackage{amsmath}
\usepackage{amsthm}
\usepackage{booktabs}
\usepackage{graphicx}
\usepackage{balance}
\usepackage{cleveref}          % For consistent cross-references
\usepackage{algorithm}
\usepackage[noend]{algorithmic}
\usepackage{xcolor}

Page Limits

Section	Limit
Main paper (submission)	8 pages
References	No limit
Appendix / Supplementary	Permitted (not counted)

ACM MM enforces a strict 8-page limit for the main body. References and supplementary materials do not count toward this limit, but reviewers are not required to read supplementary materials.

Anonymity Requirements

Double-blind review with anonymous class option:

1. No author names or affiliations in submission
2. Third-person self-citations: "Smith et al. showed..." not "we showed..."
3. Anonymize all URLs, project pages, GitHub links
4. Clear PDF metadata
5. Remove acknowledgments
6. Anonymize any embedded video/animation content

Camera-Ready Differences

After acceptance:

1. Remove review and anonymous options
2. Fill in \acmConference, \acmISBN, \acmDOI
3. Restore all author names and affiliations
4. Add appropriate \setcopyright mode
5. Enable page numbers with \settopmatter{printfolios=true}

Section Organization

ACM MM papers follow a multimedia research paper structure:

1. Abstract — Concise summary of contribution
2. Introduction — Problem, multimedia challenge, contributions (enumerate explicitly)
3. Background & Related Work — Multimedia foundations, prior methods
4. Problem Definition — Formal definition with notation
5. Proposed Approach — Core algorithm, model, or system
6. Experiments — Multimedia datasets, metrics, baselines, results
7. Discussion — Limitations, extensions, implications
8. Conclusion
9. References
10. Appendix / Supplementary (optional)

Multimedia Evaluation

ACM MM evaluation requires multimedia-specific benchmarks:

\section{Experiments}
\subsection{Datasets}
We evaluate on three standard multimedia benchmarks:

\begin{itemize}
  \item \textbf{VideoQA~\citep{tang2018movieqa}}: 6,494 questions on 308 movies
  \item \textbf{MSR-VTT~\citep{xu2016msr}}: 10,000 video-caption pairs
  \item \textbf{AudioCaps~\citep{kim2019audiocaps}}: 10,000 audio clips with captions
\end{itemize}

\subsection{Metrics}
We report standard metrics for each task:
\begin{itemize}
  \item VideoQA: accuracy
  \item Video captioning: BLEU@4, METEOR, CIDEr, SPICE
  \item Audio captioning: BLEU@4, SPIDEr
\end{itemize}

\subsection{Baselines}
We compare against state-of-the-art multimodal methods:
LSTM encoders~\citep{hodosh2013framing}, attention-based models~\citep{xu2016msa},
and vision-language transformers~\citep{radford2021learning}.

Multi-Modal Figures

ACM MM papers often contain rich multi-modal figures combining images, audio waveforms, text:

\begin{figure}[t]
  \centering
  \includegraphics[width=\linewidth]{figs/pipeline}
  \caption{Pipeline of our multimodal fusion approach.
    Video frames are encoded by a 3D CNN (green), audio spectrograms
    by a CNN (blue), and text by a BERT encoder (orange).
    Cross-modal attention (red) learns to align representations across
    modalities. Final predictions are generated by a multimodal decoder.}
  \label{fig:pipeline}
\end{figure}

% For combined audio-visual figures:
\begin{figure}[t]
  \centering
  \begin{minipage}{0.48\linewidth}
    \centering
    \includegraphics[width=\linewidth]{figs/frames}
    \subcaption{Key video frames}
  \end{minipage}
  \hfill
  \begin{minipage}{0.48\linewidth}
    \centering
    \includegraphics[width=\linewidth]{figs/spectrogram}
    \subcaption{Corresponding audio spectrogram}
  \end{minipage}
  \caption{Video and audio synchronization in our dataset.
    The spectrogram peaks (arrows) align with visual action onsets.}
  \label{fig:multimodal}
\end{figure}

References (natbib)

\bibliographystyle{ACM-Reference-Format}
\bibliography{references}

\citet{radford2021learning}  % Radford et al. (2021)
\citep{radford2021learning}  % (Radford et al. 2021)

Formatting Rules

• Format: ACM sigconf (two-column, single-spaced)
• Paper size: US Letter
• Body font: 9pt minimum, Times New Roman
• References: 8pt, unlimited pages
• Margins: Top/bottom 1in, sides 0.75in, column gap 0.25in
• Supplementary materials: Uploaded separately, not counted toward page limit

Submission Checklist

• 8 pages or fewer (main body)
• \documentclass[sigconf, review, anonymous]
• All author identification removed
• Third-person self-citations only
• Anonymized all URLs and links
• Multi-modal datasets and metrics used
• High-quality, grayscale-legible figures
• References list all authors

Camera-Ready Checklist

• Remove review and anonymous options
• Add conference metadata
• Restore author information
• Enable page numbers
• Add acknowledgments if desired