README.md

metadata

library_name: diffusers
license: cc-by-nc-4.0
pipeline_tag: image-to-image
tags:
  - video-to-video
  - video-object-removal
  - video-inpainting
  - cvpr

EffectErase: Joint Video Object Removal
and Insertion for High-Quality Effect Erasing

CVPR 2026

Yang Fu  ·  Yike Zheng  ·  Ziyun Dai  ·  Henghui Ding†

Institute of Big Data, College of Computer Science and Artificial Intelligence, Fudan University, China
† Corresponding author

This repository provides the checkpoint EffectErase.ckpt for EffectErase, as presented in the paper EffectErase: Joint Video Object Removal and Insertion for High-Quality Effect Erasing.

Abstract

Video object removal aims to eliminate dynamic target objects and their visual effects, such as deformation, shadows, and reflections, while restoring seamless backgrounds. Current methods often struggle to erase these effects and synthesize coherent backgrounds. To address this, we introduce VOR (Video Object Removal), a large-scale dataset of 60K high-quality video pairs covering various object effects. Building on VOR, we propose EffectErase, an effect-aware video object removal method that treats video object insertion as a reciprocal learning task. The model includes task-aware region guidance and an insertion-removal consistency objective to ensure high-quality video object effect erasing across diverse scenarios.

Quick Start

1. Setup repository and environment

   git clone https://github.com/FudanCVL/EffectErase.git
   cd EffectErase
   pip install -e .
   

2. Download weights

   hf download alibaba-pai/Wan2.1-Fun-1.3B-InP --local-dir Wan-AI/Wan2.1-Fun-1.3B-InP
   hf download FudanCVL/EffectErase EffectErase.ckpt --local-dir ./
   

3. Run the script

   bash script/test_remove.sh
   

   You can edit script/test_remove.sh and change these paths to use your own data:

   • --fg_bg_path: Path to the input video.
   • --mask_path: Path to the mask video (e.g., generated by SAM2.1).
   • --output_path: Path for the saved results.

BibTeX

@inproceedings{fu2026effecterase,
  title={EffectErase: Joint Video Object Removal and Insertion for High-Quality Effect Erasing},
  author={Fu, Yang and Zheng, Yike and Dai, Ziyun and Ding, Henghui},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2026}
}

Contact

If you have any questions, please feel free to reach out at aleeyanger@gmail.com.