Skip to content
@eventshift-seg

eventshift-seg

EventShift

Reliability-Aware Residual Adaptation for Day–Night RGB–Event Semantic Segmentation

Project Page · Paper · Code

EventShift day and night semantic segmentation overview

EventShift preserves RGB as the semantic anchor and introduces event information as a reliability-aware, bounded residual correction. The method improves matched-backbone nighttime performance by +4.58 mIoU while changing daytime performance by only −0.16 mIoU.

The project page contains the final paper, full controlled ablation, DSEC11 evaluation, official challenge context, qualitative comparisons, and a transparent discussion of scope and limitations.

The official implementation, environment specification, training and inference entry points, data preparation tools, and citation metadata are available in eventshift-seg/EventShift. Checkpoint downloads are being packaged separately.

Authors: Shao-Kai Liu, Yun-Tze Tsai, Chia-Yu Lin, Chia-Ming Lee, and Chih-Chung Hsu
Affiliations: National Yang Ming Chiao Tung University and National Cheng Kung University, Taiwan

Popular repositories Loading

  1. eventshift-seg.github.io eventshift-seg.github.io Public

    Project page for EventShift: reliability-aware day-night RGB-event semantic segmentation

    HTML

  2. .github .github Public

    Public profile and community metadata for the EventShift project

  3. EventShift EventShift Public

    Official implementation of EventShift for day-night RGB-event semantic segmentation

    Python

Repositories

Showing 3 of 3 repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…