An End-To-End Pipeline for Virtual Banner Replacement in Football Broadcasts
Bibliographic record
Abstract
Augmented reality has been used in sports broadcasting since the 1990s to enhance viewer engagement through virtual overlays. A key application is virtual advertising, which replaces physical advertisement banners with dynamic digital content, enabling targeted and regionspecific advertisements. This technology optimizes advertising space and increases monetization opportunities for broadcasters. However, traditional augmented reality solutions require specialized hardware, such as instrumented cameras and virtual-ready LED panels, along with manual calibration and prior environmental knowledge. These constraints make its implementation costly and less adaptative. In this work, we propose a first fully automated end-to-end pipeline that seamlessly integrates augmented reality advertising into sports broadcasts using only the main camera feed. Our approach leverages state-of-the-art deep neural networks to identify the advertisement banner, estimate camera motion, and dynamically composite virtual content without additional hardware or manual intervention. We validate our pipeline on football broadcasts using our novel SoccerNet-banner dataset, the first dataset for training and evaluating banner segmentation models, and demonstrate high-quality virtual banner replacement on SoccerNet videos. Therefore, our pipeline unlocks new possibilities for personalized content and advances AI-powered sports broadcasting by eliminating hardware dependencies and manual calibration. Our code and dataset are available at https://github.com/SoccerNet/snbanner.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.009 | 0.008 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".