Merge pull request #16 from MehdiH7/exposureEngine

Author: stevenah

README.md

@@ -23,6 +23,7 @@ Currently, we have published the following datasets:
 **Sport Datasets**
 * Alfheim, Soccer video and player position dataset. [ [publication](https://dl.acm.org/doi/10.1145/2557642.2563677) ]
 * ARX, A Text-Classification Dataset Consisting of Norwegian Soccer Articles from VG and TV2. [ [publication](https://ieeexplore.ieee.org/abstract/document/8877417/) ]
+* ExposureEngine, Oriented Logo Detection & Sponsor Visibility Analytics (Dataset).
 * Heimdallr, A Dataset For Sport Analysis.
 * ScopeSense, A 8.5-month sport, nutrition, and lifestyle lifelogging dataset.
 * Soccer Summarization, Soccer game captions and summary in English for game summarization. [ [publication](https://dl.acm.org/doi/10.1145/3552463.3557019) ]

datasets/exposure-engine.md

@@ -0,0 +1,53 @@
+---
+title: 'ExposureEngine'
+desc: 'Oriented Logo Detection and Sponsor Visibility Analytics in Sports Broadcasts'
+thumbnail: /thumbnails/exposure-engine.jpg
+publication: ---
+github: ---
+hidden: false
+tags:
+  - Logo Detection
+  - Oriented Object Detection
+  - Analytics
+---
+
+# ExposureEngine: Oriented Logo Detection & Sponsor Visibility Analytics (Dataset)
+
+[![Dataset](https://img.shields.io/badge/HuggingFace-Dataset-blue)](https://huggingface.co/datasets/SimulaMet-HOST/ExposureEngine)
+
+Rotation-aware **OBB** annotations for sponsor logos in professional soccer broadcasts — built for **sports analytics**, **YOLO OBB training**, and **sponsorship measurement**.
+
+**ExposureEngine** provides high-quality *oriented bounding box (OBB)* polygon annotations of sponsor logos from professional soccer footage to enable accurate **sponsor visibility measurement**, **logo tracking**, and **rotation-aware detection**.
+
+---
+
+## 📂 Dataset Overview
+
+- **Source:** 97 broadcast highlight clips from the *2024 Swedish Elite Soccer League*
+- **Frames:** 1,103 curated frames (sampled at 1 FPS)
+- **Annotations:** 670 unique sponsor logos with **OBB polygon coordinates** (YOLO format)
+- **Coverage:** 32 matches · 16 teams · diverse events: Goals, Shots, Yellow Cards, Offsides, Substitutions
+- **Splits:** Train / Validation / Test = **80 / 10 / 10**
+
+---
+
+## 🔍 Why Oriented Bounding Boxes?
+
+OBB captures **rotation**, **skew**, and **perspective** — common in sports due to dynamic camera angles, jersey folds, and slanted boards. This improves **coverage accuracy** and **downstream analytics** over traditional HBB.
+
+---
+
+## 🚀 Applications
+
+- Sports **sponsorship analytics** & reporting
+- **YOLO-based** OBB training & benchmarking
+- Brand **visibility measurement** in broadcasts
+- Content adaptation for **social media** & OTT
+
+---
+
+## 📩 Contact
+For inquiries, collaborations, or questions:
+- 📧 Mehdi Houshmand: [mehdi@forzasys.com](mailto:mehdi@forzasys.com)
+- 📧 Cise Midoglu: [cise@forzasys.com](mailto:cise@forzasys.com)
+- 📧 Pål Halvorsen: [paalh@simula.no](mailto:paalh@simula.no)