Summary
EffectErase is an effect-aware video object removal method built on VOR, a newly introduced large-scale dataset containing 60K high-quality video pairs for video object removal. The approach treats video object insertion as an inverse auxiliary task within a reciprocal learning scheme, allowing removal and insertion to reinforce each other. Authored by Yang Fu, Yike Zheng, and Ziyun Dai in the computer vision field, the work was released on arXiv (2503.16887) in March 2026. This forum post provides a brief Chinese summary of the paper's abstract alongside the original English abstract, covering the dataset contribution and the reciprocal-learning method design. It is a concise reference for researchers tracking video editing, object removal, and video object insertion research.
Paper Overview
Field: Computer Vision (CV)
Authors: Yang Fu, Yike Zheng, Ziyun Dai
Release date: 2026-03-19
arXiv: 2503.16887
Abstract (English)
We introduce VOR, a large-scale dataset with 60K high-quality video pairs for video object removal. Building on VOR, we propose EffectErase, an effect-aware video object removal method that treats video object insertion as the inverse auxiliary task within a reciprocal learning scheme.
Key Contributions
- VOR dataset: A large-scale dataset of 60K high-quality video pairs designed for video object removal.
- EffectErase method: An effect-aware approach to video object removal.
- Reciprocal learning scheme: Video object insertion is framed as the inverse auxiliary task, mutually benefiting the removal task.
---
*Auto-collected on 2026-03-22.*
This page is an English static mirror generated for search and AI citation.
It may be a full translation or structured summary of the Chinese original.
Canonical interactive discussion lives on the Chinese page:
https://zhichai.net/topic/177168987