LVVO

The Lecture Video Visual Objects (LVVO) dataset is a benchmark designed for object detection in lecture video frames. It provides high-quality annotations of visual content such as tables, charts, images, and illustrations in real university lecture recordings. Provide:

📚 Dataset Overview

• Total Images: 4,000 unique frames extracted from lecture videos
• Manually Annotated Subset (LVVO 1k): 1,000 frames
• Automatically Labeled Subset (LVVO 3k): 3,000 frames
• Source: Lecture recordings from videopoints.org, covering 8 instructors across 13 courses and 3 domains (Biology, Computer Science, Geoscience