VACaps: A Domain-Specific Image-Caption Dataset for Video Analytics
Abstract
This is an image–caption dataset comprising 30k images, 210k captions, and 304k bounding box annotations, designed to support Video Analytics Research and Applications. It contains real-world images depicting activities and events relevant to surveillance, public safety, and abnormal behavior detection. Contents: 30,000 images 210,000 captions (seven captions per image) 304,043 bounding box annotations (multiple bounding boxes per image) File Formats: Images: PNG / JPG / JPEG Bounding Boxes: TXT Captions: CSV The dataset is intended for: Vision–Language Modeling Real-time Object Detection Model Development Text-based Image Retrieval for Video Analytics Multimodal Reasoning
Fields of Study & Subjects
Multimodal Machine Learning Applications
Image Retrieval and Classification Techniques
Advanced Neural Network Applications
Publication details
- Journal / source
- Zenodo (CERN European Organization for Nuclear Research)
- Publication date
- 4 March 2026
- Language
- EN
- Document type
- Dataset