Research map / Detection and segmentation / Open-set and few-shot detection
Adversarial Scene Editing: Automatic Object Removal from Weak Supervision
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add a mask prior using unpaired Pascal segmentation masks (ablation) | Adversarial Scene Editing · COCO | false removal: 36 -> 16 | Helped |
| Add a mask prior using unpaired Pascal segmentation masks (ablation) | Adversarial Scene Editing · COCO | pSNR: 19.97 -> 22.64 | Helped |
| Add a mask prior using unpaired Pascal segmentation masks (ablation) | Adversarial Scene Editing · COCO | removal success (all): 94 -> 73 | Hurt |
| Increase the Pascal mask prior from 10 masks to more samples (ablation) | Adversarial Scene Editing · COCO person removal | removal success: 59 -> 81 | Helped |
| Dilate Mask R-CNN masks before in-painting | Mask R-CNN removal baseline · COCO | removal success (all): 68 -> 75 | Helped |
| Jointly optimize the mask generator and in-painter instead of freezing the in-painter (ablation) | Adversarial Scene Editing · COCO | removal success: 0.68 vs 0.73 | Helped |
| Jointly optimize the mask generator and in-painter instead of freezing the in-painter (ablation) | Adversarial Scene Editing · COCO | mIoU: 0.19 vs 0.22 | Helped |
| Jointly optimize the mask generator and in-painter instead of freezing the in-painter (ablation) | Adversarial Scene Editing · COCO | perceptual loss: 0.10 vs 0.08 | Helped |
| Add the mask buffer for reconstruction training on randomly paired previous-batch masks (ablation) | In-painter · COCO object removal | in-painting performance | Helped |
| Use local pixel-level rather than global real/fake loss (ablation) | In-painter · COCO object removal | perceptual loss: 0.12 vs 0.10 | Helped |
| Use local pixel-level rather than global real/fake loss (ablation) | In-painter · COCO object removal | pSNR | Hurt |
| Add total variation and style losses (ablation) | In-painter · COCO object removal | pSNR | Helped |
Closest papers by meaning
- Neural Photo Editing with Introspective Adversarial Networks (ICLR 2017 · similarity 0.88)
- PandA: Unsupervised Learning of Parts and Appearances in the Feature Maps of GANs (ICLR 2023 · similarity 0.87)
- MiniMax-Remover: Taming Bad Noise Helps Video Object Removal (NeurIPS 2025 · similarity 0.86)
- E$^2$GAN: Efficient Training of Efficient GANs for Image-to-Image Translation (ICML 2024 · similarity 0.85)
- CLIPAway: Harmonizing focused embeddings for removing objects via diffusion models (NeurIPS 2024 · similarity 0.85)
- Paint by Inpaint: Learning to Add Image Objects by Removing Them First (CVPR 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.