AdSpark offers large dataset and benchmark for ad video creation
AdSpark: A Large-Scale Dataset and Benchmark for Product-Centric Advertisement Video Generation
Computer Vision and Pattern Recognition
Summary
Creating good promotional videos that clearly show a product and its selling points is hard, especially because there isn’t enough data or ways to check if the videos are good. The authors created AdSpark, a very large collection of pictures, text prompts, and videos about products. They also made a test to measure how well generated videos keep the product clear, follow instructions, and tell a good story. Their tests show current video generation models still struggle with keeping product details and storytelling over many shots.
What this means in practice
- •For advertising technology teams: Train and evaluate AI models that generate multi-shot product advertisement videos with clear storytelling and selling points.$Commercial implications: Enables creating automated promotional video generation tools for e-commerce platforms, improving product marketing at scale.
- •For content generation platform developers: Integrate product-specific video creation capabilities that preserve brand identity and align with creative advertising scripts.
Authors
Zhifei Yang, Zhao Jiang, Keyang Lu, Honghe Zhu, Zheng Zhang, Jingjing Lv, Changping Peng, Ching Law, Zhen Xiao
Abstract
Product-centric advertisement video generation aims to create promotional videos that preserve fine-grained product identity while presenting selling points through coherent multi-shot narratives. However, this emerging task remains underexplored due to the lack of large-scale advertisement-specific datasets and comprehensive evaluation frameworks. To address this gap, we introduce \textbf{AdSpark}, a large-scale dataset and benchmark for product-centric advertisement video generation, based on data from a major e-commerce platform. \textit{AdSpark-300K} contains approximately 300K reference image--prompt--video triplets, comprising a real-world subset and a synthetic subset. Each sample provides structured advertisement annotations, including product identity annotations, selling-point descriptions, creative plans, and aligned audio scripts, enabling models to learn product preservation and advertisement-oriented visual storytelling. We further propose \textit{AdSpark-Bench}, a diagnostic benchmark that evaluates generated advertisements across six dimensions, including visual quality, product fidelity, instruction adherence, temporal coherence, audio alignment, and advertisement effectiveness. Based on AdSpark-Bench, we evaluate representative models, revealing key challenges in product preservation, multi-shot storytelling, and selling-point visualization. Experiments with AdSpark-300K-finetuned models further validate the effectiveness of our dataset. AdSpark provides a unified dataset and benchmark for future research, and we will release the dataset upon acceptance.