Event Quality Score (EQS): Assessing the Realism of Simulated Event Camera Streams via Distances in Latent Space
Authors:
- 11 June 2025
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 5105-5113
Abstract
Event cameras promise a paradigm shift in vision sensing with their low latency, high dynamic range, and asynchronous nature of events. Unfortunately, the scarcity of high-quality labeled datasets hinders their widespread adoption in deep learning-driven computer vision. To mitigate this, several simulators have been proposed to generate synthetic event data for training models for detection and estimation tasks. However, the fundamentally different sensor design of event cameras compared to traditional frame-based cameras poses a challenge for accurate simulation. As a result, most simulated data fail to mimic data captured by real event cameras. Inspired by existing work on using deep features for image comparison, we introduce event quality score ($E Q S$), a quality metric that utilizes activations of the RVT architec-Keywords
All Related Versions
This publication has 37 references indexed in Scilit:
- Physical-Based Event Camera Simulator2024
- Event-Based Sensing for Improved Traffic Detection and Tracking in Intelligent Transport Systems Toward Sustainable Mobility2024
- Neuromorphic high-frequency 3D dancing pose estimation in dynamic environmentNeurocomputing, 2023
- DVS-Voltmeter: Stochastic Process-Based Event Simulator for Dynamic Vision Sensors2022
- Reducing the Sim-to-Real Gap for Event Cameras2020
- Events-To-Video: Bringing Modern Computer Vision to Event Cameras2019
- HATS: Histograms of Averaged Time Surfaces for Robust Event-Based Object Classification2018
- The Unreasonable Effectiveness of Deep Features as a Perceptual Metric2018
- Vertical landing for micro air vehicles using event‐based optical flowJournal of Field Robotics, 2017
- U-Net: Convolutional Networks for Biomedical Image Segmentation2015