AD-MIR: Bridging the Gap from Perception to Persuasion in Advertising Video Understanding via Structured Reasoning

Multimodal understanding of advertising videos is essential for interpreting the intricate relationship between visual storytelling and abstract persuasion strategies. However, despite excelling at general search, existing agents often struggle to bridge the cognitive gap between pixel-level perception and high-level marketing logic. To address this challenge, we introduce AD-MIR, a framework designed to decode advertising intent via a two-stage architecture. First, in the Structure-Aware Memory Construction phase, the system converts raw video into a structured database by integrating semantic retrieval with exact keyword matching. This approach prioritizes fine-grained brand details, such as logos and on-screen text, while dynamically filtering out irrelevant background noise to isolate key protagonists. Second, the Structured Reasoning Agent mimics a marketing expert through an iterative inquiry loop, decomposing the narrative to deduce implicit persuasion tactics. Crucially, it employs an evidence-based self-correction mechanism that rigorously validates these insights against specific video frames, automatically backtracking when visual support is lacking. Evaluation on the AdsQA benchmark demonstrates that AD-MIR achieves state-of-the-art performance, surpassing the strongest general-purpose agent, DVD, by 1.8 and 9.5 percentage points in strict and relaxed accuracy, respectively. These results underscore that effective advertising understanding demands explicitly grounding abstract marketing strategies in pixel-level evidence. The code is available at https://github.com/Little-Fridge/AD-MIR.

Many online videos are advertisements, and understanding them often requires more than recognizing objects or reading text on the screen. A person may need to infer what feeling the ad is trying to create, what audience it is targeting, or what message it is trying to persuade viewers to believe. Current AI systems can often describe what appears in a video, but they may guess the deeper advertising intent without checking whether the video actually supports that guess. This paper introduces AD-MIR, an AI system designed to answer questions about advertising videos in a more evidence-based way. Instead of giving an answer immediately, AD-MIR first organizes useful information from the video, such as key scenes, spoken words, on-screen text, brand details, and important people or objects. It then reasons about the ad like a marketing analyst, but repeatedly checks its ideas against specific video frames. If the evidence is missing or weak, the system revises its answer. On a standard test of advertising video understanding, AD-MIR answers more accurately than previous strong systems. More importantly, it makes the reasoning process easier to inspect, because its answers are tied to concrete visual evidence rather than unsupported guesses.