Year: 2025
Type: article
Abstract: We propose an Event-Based Snow Removal algorithm called EBSnoR. We developed a technique to measure the dwell time of snowflakes on a pixel using event-based camera data, which is used to carry out a statistically optimal dwell time thresholding to partition event stream into snowflake and background events. The effectiveness of the proposed EBSnoR was verified qualitatively on a new dataset called UDayton25EBSnow comprised of front-facing event-based camera in a car driving through snow with manually annotated bounding boxes around surrounding vehicles, as well as a quantitatively using new snowflake event simulator called EBSnoGen. Qualitatively, EBSnoR correctly identifies events corresponding to snowflakes; and quantitatively, EBSnoR showed accuracy of 96.19%. Additional experiments showed that snow removal improved event-based object detection performance.
Authors: Abigail Wolf, Osama Alsattam, Shannon Brooks-Lehnert, Keigo Hirakawa
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funder: Ford Motor Company
OpenAlex ID: W4413785594