← Back to the feed
Energy & materialsResearch findings

SUMMA: A Multimodal Model for Commercially Valuable Ad Summarization

Original published 2025.08.28 18:19 KSTAdded to Blackdeck 2026.10.23 02:42 KST

What has changed?

A multimodal model that automatically summarizes video ads by focusing on commercially valuable content.

Key points

  1. We propose SUMMA, a multimodal model for summarizing video ads with commercial value.
  2. It integrates video frames and text transcripts to generate explainable, high-value summaries.
  3. Offline and online experiments show a 1.5% increase in ad revenue over baselines.

Connect it to your work

Leverages video frames and ASR/OCR transcripts to generate explainable, commercially relevant summaries.

What remains unproven

This research note is based on an abstract; check the original for test conditions and application scope.

Originals used for this note

Links open the original publisher in a new tab.

  1. arxivSUMMA: A Multimodal Large Language Model for Advertisement Summarization