Exploring AI Applications in iQIYI Video Advertising
The talk explains how iQIYI leverages AI—through video understanding, tag generation, graph construction, and scene‑based point selection—to automatically create and place diverse ad formats, optimizing ad inventory, relevance, and revenue while meeting brand and production constraints.
Speaker Liu Qiyue, a scientist at iQIYI, introduces the use of artificial intelligence in video advertising on the iQIYI platform. The talk focuses on how AI can understand and process video content and advertising assets, creating value for ad inventory and revenue.
Background: Advertising aims to drive consumption, ranging from long‑term brand ads to short‑term performance ads. Effective advertising requires matching user demand with appropriate ad placement, often through scene‑based marketing.
Scene Generation / Recommendation: AI helps identify suitable ad slots ("points") within video scenes, generate or recommend these points, and assist in creating ad materials. Examples include inserting beverage ads in romantic scenes or payment ads during checkout moments.
Video Understanding: Video analysis involves recognizing objects, events, and scenes to form a rich set of tags. These tags can be commercial (valuable for advertisers) or creative (useful for content creators). Combining low‑level tags with relational analysis improves tag precision and relevance.
Video Graph: By linking tags across objects, events, and scenes, a video graph is built, enabling deeper semantic understanding and supporting complex advertiser requirements.
Ad Point Selection Process (VideoIn): The workflow considers brand requirements (visibility, non‑obstruction, co‑presence with stars), production constraints (no conflict with competing ads), and technical feasibility (ability to embed ads). The goal is to select points that satisfy both commercial and creative criteria.
Ad Formats: Various formats are discussed, including pre‑roll, mid‑roll, post‑roll, overlay ads ("stickers"), and product placement (both pre‑ and post‑production). Each format has different scalability and implementation challenges.
AI‑Assisted Video Marketing: AI assists in generating scene‑based ad points, creating pre‑recap summaries, selecting cover images, and generating titles. It also helps evaluate the feasibility, richness, and timeliness of ad placements.
Overall, the presentation outlines how AI techniques—such as object detection, event recognition, scene classification, and knowledge graph construction—enable scalable, context‑aware advertising on video platforms.
iQIYI Technical Product Team
The technical product team of iQIYI
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.