RibbonNet: A Curated Dataset of ID Card Ribbons for Institutional Role Classification
Description
Data Description RibbonNet is a curated image dataset developed for the task of ID card ribbon detection and institutional role classification in educational environments. The dataset consists of 550 raw images and 3,343 annotated images, capturing individuals wearing ID card ribbons corresponding to three categories: Admin, Teacher, and Student. These ribbons serve as visual identifiers in many academic institutions. Images were acquired using DSLR cameras, smartphones (iPhone 14 Pro, Xiaomi), and sourced from publicly available social media content. The collection was conducted at Daffodil Smart City, Dhaka, Bangladesh, between September 2024 and February 2025, under various lighting conditions and viewing angles to ensure dataset diversity and real-world applicability.
Files
Steps to reproduce
The dataset includes: 550 raw images (Admin: 155, Student: 160, Teacher: 235) 3,343 augmented and annotated images in YOLOv8 format Manual annotations created using Roboflow, focusing on bounding boxes around ID card ribbons only Augmentation techniques applied: brightness adjustment, flipping, rotation, and blurring Each image is paired with a .txt file containing YOLOv8-style annotations, specifying the class (0 = Admin, 1 = Student, 2 = Teacher) and normalized bounding box coordinates. The dataset is organized into two main directories: Raw Data: Contains class-wise folders for Admin, Teacher, and Student Augmented Data: Contains Images/ and Labels/ folders with corresponding YOLOv8 annotation files This dataset supports research and development in privacy-aware, role-based identification systems, especially in surveillance, access control, and smart campus automation where full-face or ID scans are intrusive or impractical. Data Format: Images: JPG Annotations: YOLOv8-compatible .txt files Potential Use Cases: 1. Training and evaluating YOLO-based object detection models 2. Real-time role recognition systems in educational institutions 3. Privacy-conscious visual identification methods 4. Research in smart surveillance and access control systems