MultiviewX: Synchronized Multi-Camera Pedestrian Detection from ECCV 2020

MultiviewX is a synthetic dataset for multiview pedestrian detection introduced by hou-yz for the ECCV 2020 paper "Multiview Detection with Feature Perspective Transformation." It provides synchronized camera views and ground truth annotations to support research in feature perspective transformation and 3D occupancy mapping.

Use Cases

• Training pedestrian detectors using synchronized multiview image frames
• Developing feature perspective transformation models using camera calibration matrices
• Estimating 3D ground plane occupancy from multiple 2D camera inputs

Strengths

• Peer-reviewed ECCV 2020 publication origin
• Provides synchronized multiview image sets
• Includes camera calibration parameters for perspective transformation

Limitations

• Synthetic data origin may lack real-world visual noise and lighting complexities
• Task-specific focus on pedestrian detection limits broader object detection utility

Provenance

Source

ECCV 2020 paper "Multiview Detection with Feature Perspective Transformation"

Collection Method

Synthetic generation

Freshness

Last updated April 2025; based on 2020 research publication.