pith. sign in

arxiv: 2607.00920 · v2 · pith:WS465DT3new · submitted 2026-07-01 · 💻 cs.CV

GMO-E²DIT: Grounded Multi-Operation Editing for E-Commerce Images

Zipeng Guo , Xiaoan Liu , Lichen Ma , Cheng Wang , Yu He , Xiaolong Fu , Jingling Fu , Xinyuan Shan
show 4 more authors
Shaojie Guo Luohang Liu Junshi Huang Yan Li
This is my paper

Pith reviewed 2026-07-03 21:21 UTC · model grok-4.3

classification 💻 cs.CV
keywords image editinge-commerce imagesvision-language modelsmulti-operation editingagentic frameworksmask-conditioned generationreflection-driven executiongrounded editing
0
0 comments X

The pith

A VLM agent builds region-grounded edit agendas that a mask editor follows step by step to handle multiple precise changes on e-commerce images.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents GMO-E²DIT as an agentic framework that splits the editing process into planning and execution phases. A vision-language model first turns vague instructions into a sequence of localized operations with masks, then a dedicated editor applies them while a reflection loop checks results and adjusts as needed. This setup targets real commercial needs where edits must be accurate, partial progress must be kept, and errors fixed without global changes. The authors also release a data pipeline for training and EComEditBench for testing such systems. If the separation works, it reduces the failures common in one-step models that try to resolve intent, location, and synthesis all at once.

Core claim

GMO-E²DIT couples a VLM agent that constructs a region-grounded edit agenda with a mask-conditioned image editor and a reflection-driven loop; this decoupling of cognitive reasoning from generative rendering allows reliable execution of multiple localized operations on e-commerce images even from underspecified instructions, while preserving unmodified content and recovering from partial failures.

What carries the argument

The VLM-constructed region-grounded edit agenda, executed through operation-aware masks and a reflection loop that inspects intermediate outputs to decide next steps.

If this is right

  • The framework preserves safe partial progress across iterative edits and retries unfinished operations.
  • It recovers from errors through the reflection loop instead of failing the entire task.
  • A unified data pipeline supplies aligned supervision for planning, execution, and reflection stages.
  • The system yields higher instruction accuracy and edit fidelity than prior baselines on structured multi-operation tasks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The planning-execution split could extend to other multi-step visual tasks such as video editing or 3D scene modification.
  • EComEditBench offers a reusable testbed for evaluating grounded instruction following beyond e-commerce.
  • Similar agentic loops might improve reliability in any domain where instructions are vague but spatial precision is required.

Load-bearing premise

The vision-language model can consistently turn ambiguous instructions into correct region-specific edit plans that the downstream editor can follow without major grounding errors.

What would settle it

A collection of underspecified editing instructions on which the VLM agent produces agendas that misidentify regions or operations, causing the mask editor to apply changes to the wrong areas or skip required steps even after reflection attempts.

Figures

Figures reproduced from arXiv: 2607.00920 by Cheng Wang, Jingling Fu, Junshi Huang, Lichen Ma, Luohang Liu, Shaojie Guo, Xiaoan Liu, Xiaolong Fu, Xinyuan Shan, Yan Li, Yu He, Zipeng Guo.

Figure 1
Figure 1. Figure 1: Qualitative demonstration. (a) Single-task editing operations. (b) Complex multi-task and [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of our framework. Training involves three stages: 1) SFT and rule-based GRPO [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overview of the data synthesis pipeline. Stage 1 extracts structured visual and textual assets, [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Overview of training and test data. (a) Sample counts for single-task and multi-task in￾structions in training. (b) Evaluation set composi￾tion by instruction complexity, highlighting bench￾mark diversity. Unified Synthesis of Image Editing Data. We synthesize training samples by integrating three core components: the raw product im￾age, a generatively-inpainted background, and structured elements extracte… view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative comparison on image editing tasks. Highlighted regions indicate areas where [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: User study between our method and closed-source models regarding In￾struction Following and Consistency [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 6
Figure 6. Figure 6: User study between our method and closed-source models regarding In￾struction Following and Consistency [PITH_FULL_IMAGE:figures/full_fig_p009_6.png] view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative comparison on image editing tasks. Highlighted regions indicate areas where [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: More visual comparison with closed-source methods across multiple tasks. [PITH_FULL_IMAGE:figures/full_fig_p021_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Comparison of single-turn vs. multi-turn instruction execution. (Example 1) [PITH_FULL_IMAGE:figures/full_fig_p022_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Comparison of single-turn vs. multi-turn instruction execution. (Example 2) [PITH_FULL_IMAGE:figures/full_fig_p022_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Comparison of single-turn vs. multi-turn instruction execution. (Example 3) [PITH_FULL_IMAGE:figures/full_fig_p023_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Comparison of single-turn vs. multi-turn instruction execution. (Example 4) [PITH_FULL_IMAGE:figures/full_fig_p023_11.png] view at source ↗
read the original abstract

Real-world e-commerce image editing often requires multiple, localized, and auditable operations rather than global restyling. This compositional nature poses a dual challenge: models must precisely apply all requested edits to the correct regions while preserving unmodified content, even under ambiguous instructions. Existing one-shot editors conflate intent resolution, spatial grounding, and synthesis into a single step, frequently resulting in partial execution failures, which is unacceptable for commercial scenarios. To address this, we introduce GMO-E$^2$DIT, an agentic editing framework that couples a Vision-Language Model (VLM) with a mask-conditioned image editor to tackle structured multi-turn task completion. Given an underspecified instruction, the VLM agent constructs a region-grounded edit agenda, effectively decoupling cognitive reasoning from generative rendering. The framework then executes sub-programs via operation-aware masks and references, utilizing a reflection-driven loop to inspect intermediate results and determine the subsequent state. This iterative mechanism reliably preserves safe partial progress, retries unfinished operations, and recovers from errors. Furthermore, we develop a unified data pipeline providing aligned supervision for planning, execution, and reflection, alongside EComEditBench, a comprehensive benchmark for instruction-driven evaluation. Extensive experiments demonstrate that GMO-E$^2$DIT achieves competitive performance compared to strong closed-source models, yielding superior instruction accuracy and edit fidelity over existing baselines.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper introduces GMO-E²DIT, an agentic framework coupling a VLM with a mask-conditioned image editor for multi-operation e-commerce image editing. Given underspecified instructions, the VLM builds a region-grounded edit agenda to decouple reasoning from rendering; sub-programs are executed via operation-aware masks and references, with a reflection loop for inspecting results, preserving partial progress, and recovering from errors. The work also contributes a unified data pipeline for planning/execution/reflection supervision and EComEditBench for instruction-driven evaluation, claiming competitive performance versus closed-source models and superior instruction accuracy plus edit fidelity over baselines.

Significance. If the performance claims hold with rigorous quantitative support, the framework could offer a practical advance for commercial e-commerce editing by handling compositional, localized operations more reliably than one-shot models. The explicit separation of VLM planning from mask-conditioned synthesis, combined with the new benchmark and data pipeline, would provide reusable components for similar multi-step visual tasks.

major comments (2)
  1. [Abstract] Abstract: the central claim of 'superior instruction accuracy and edit fidelity over existing baselines' and 'competitive performance compared to strong closed-source models' is asserted without any quantitative metrics, tables, error bars, or statistical details in the provided text; this absence makes the headline result unverifiable and load-bearing for the contribution.
  2. [Abstract] Abstract: the assumption that the VLM agent reliably constructs a region-grounded edit agenda from underspecified instructions (thereby decoupling cognitive reasoning from generative rendering) is presented as a key architectural choice, yet no validation, ablation, or failure-mode analysis is supplied to show this step is robust or responsible for the claimed gains.
minor comments (1)
  1. The acronym GMO-E²DIT and its expansion are introduced without an explicit breakdown of the components (Grounded Multi-Operation Editing) that would aid readers.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for the constructive feedback. We address the two major comments on the abstract below, clarifying where quantitative support and component validation appear in the full manuscript while noting revisions to improve self-containment of the abstract.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim of 'superior instruction accuracy and edit fidelity over existing baselines' and 'competitive performance compared to strong closed-source models' is asserted without any quantitative metrics, tables, error bars, or statistical details in the provided text; this absence makes the headline result unverifiable and load-bearing for the contribution.

    Authors: We agree the abstract states the headline claims without numerical values. The full manuscript reports these results with tables, metrics (instruction accuracy, edit fidelity, FID, CLIP scores), error bars from repeated runs, and statistical comparisons in Section 4 and the associated tables. We will revise the abstract to include the key quantitative highlights (e.g., specific accuracy and fidelity deltas versus baselines and closed-source models) so the claims are verifiable from the abstract alone. revision: yes

  2. Referee: [Abstract] Abstract: the assumption that the VLM agent reliably constructs a region-grounded edit agenda from underspecified instructions (thereby decoupling cognitive reasoning from generative rendering) is presented as a key architectural choice, yet no validation, ablation, or failure-mode analysis is supplied to show this step is robust or responsible for the claimed gains.

    Authors: The abstract summarizes the design choice at a high level. The manuscript supplies the requested validation: Section 4.3 contains ablations that isolate the contribution of the region-grounded agenda construction (with and without it), and the supplementary material includes failure-mode analysis and robustness checks under ambiguous instructions. We will add a short clause to the abstract noting that this decoupling is supported by the experimental analysis in the paper. revision: partial

Circularity Check

0 steps flagged

No significant circularity; framework and claims are self-contained

full rationale

The provided abstract and context describe an architectural framework (VLM agent for planning + mask-conditioned editor + reflection loop) and a new benchmark (EComEditBench) without any equations, fitted parameters, or derivations. No self-citations are invoked as load-bearing uniqueness theorems or ansatzes. Performance claims rest on empirical comparison rather than reducing to input data by construction. The decoupling of reasoning from rendering is presented as a design choice whose value is asserted via overall metrics, not as a self-definitional or fitted-input step. This matches the default expectation of no circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

Abstract-only review yields limited visibility into parameters or axioms; framework implicitly assumes VLM planning reliability and mask precision without stated free parameters or new entities.

axioms (1)
  • domain assumption Vision-language models can reliably decompose underspecified instructions into region-grounded edit agendas
    Central to the decoupling step described in the abstract.

pith-pipeline@v0.9.1-grok · 5806 in / 1145 out tokens · 17446 ms · 2026-07-03T21:21:45.627819+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

50 extracted references · 50 canonical work pages · 12 internal anchors

  1. [1]

    Tim Brooks, Aleksander Holynski, and Alexei A. Efros. Instructpix2pix: Learning to follow image editing instructions. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 18392–18402, 2023

    work page 2023

  2. [2]

    Guid- ing instruction-based image editing via multimodal large language models

    Tsu-Jui Fu, Wenze Hu, Xianzhi Du, William Yang Wang, Yinfei Yang, and Zhe Gan. Guid- ing instruction-based image editing via multimodal large language models. InInternational Conference on Learning Representations, 2024

    work page 2024

  3. [3]

    Anyedit: Mastering unified high-quality image editing for any idea

    Qifan Yu, Wei Chow, Zhongqi Yue, Kaihang Pan, Yang Wu, Xiaoyang Wan, Juncheng Li, Siliang Tang, Hanwang Zhang, and Yueting Zhuang. Anyedit: Mastering unified high-quality image editing for any idea. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2025

    work page 2025

  4. [4]

    Qwen-Image Technical Report

    Chenfei Wu, Jiahao Li, Jingren Zhou, Junyang Lin, Kaiyuan Gao, Kun Yan, Sheng-ming Yin, Shuai Bai, Xiao Xu, Yilei Chen, et al. Qwen-image technical report.arXiv preprint arXiv:2508.02324, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  5. [5]

    FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space

    Black Forest Labs. Flux.1 kontext: Flow matching for in-context image generation and editing in latent space.arXiv preprint arXiv:2506.15742, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  6. [6]

    Insightedit: Towards better instruction following for image editing

    Yingjing Xu, Jie Kong, Jiazhi Wang, Xiao Pan, Bo Lin, and Qiang Liu. Insightedit: Towards better instruction following for image editing. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2694–2703, 2025

    work page 2025

  7. [7]

    Imageedit-r1: Boosting multi-agent image editing via reinforcement learning.arXiv preprint arXiv:2603.08059, 2026

    Yiran Zhao, Yaoqi Ye, Xiang Liu, Michael Qizhe Shieh, and Trung Bui. Imageedit-r1: Boosting multi-agent image editing via reinforcement learning.arXiv preprint arXiv:2603.08059, 2026

    work page arXiv 2026

  8. [8]

    Talk2image: A multi-agent system for multi-turn image generation and editing

    Shichao Ma, Yunhe Guo, Jiahao Su, Qihe Huang, Zhengyang Zhou, and Yang Wang. Talk2image: A multi-agent system for multi-turn image generation and editing. InProceedings of the AAAI Conference on Artificial Intelligence, volume 40, pages 32437–32445, 2026

    work page 2026

  9. [9]

    Uniworld-V2: Reinforce Image Editing with Diffusion Negative-aware Finetuning and MLLM Implicit Feedback

    Zongjian Li, Zheyuan Liu, Qihui Zhang, Bin Lin, Feize Wu, Shenghai Yuan, Zhiyuan Yan, Yang Ye, Wangbo Yu, Yuwei Niu, et al. Uniworld-v2: Reinforce image editing with diffusion negative-aware finetuning and mllm implicit feedback.arXiv preprint arXiv:2510.16888, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  10. [10]

    Enabling instructional image editing with in-context generation in large scale diffusion transformer

    Zechuan Zhang, Ji Xie, Yu Lu, Zongxin Yang, and Yi Yang. Enabling instructional image editing with in-context generation in large scale diffusion transformer. InThe Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025

    work page 2025

  11. [11]

    Meta-CoT: Enhancing Granularity and Generalization in Image Editing

    Shiyi Zhang, Yiji Cheng, Tiankai Hang, Zijin Yin, Runze He, Yu Xu, Wenxun Dai, Yunlong Lin, Chunyu Wang, Qinglin Lu, et al. Meta-cot: Enhancing granularity and generalization in image editing.arXiv preprint arXiv:2604.24625, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  12. [12]

    Chargen: High accurate character-level visual text generation model with multimodal encoder.arXiv preprint arXiv:2412.17225, 2024

    Lichen Ma, Tiezhu Yue, Pei Fu, Yujie Zhong, Kai Zhou, Xiaoming Wei, and Jie Hu. Chargen: High accurate character-level visual text generation model with multimodal encoder.arXiv preprint arXiv:2412.17225, 2024

    work page arXiv 2024

  13. [13]

    Objectmover: Generative object movement with video prior

    Xin Yu, Tianyu Wang, Soo Ye Kim, Paul Guerrero, Xi Chen, Qing Liu, Zhe Lin, and Xiaojuan Qi. Objectmover: Generative object movement with video prior. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17682–17691, 2025

    work page 2025

  14. [14]

    Towards enhanced image inpainting: Mitigating unwanted object insertion and preserving color con- sistency

    Yikai Wang, Chenjie Cao, Junqiu Yu, Ke Fan, Xiangyang Xue, and Yanwei Fu. Towards enhanced image inpainting: Mitigating unwanted object insertion and preserving color con- sistency. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 23237–23248, 2025

    work page 2025

  15. [15]

    Brushnet: A plug-and-play image inpainting model with decomposed dual-branch diffusion

    Xuan Ju, Xian Liu, Xintao Wang, Yuxuan Bian, Ying Shan, and Qiang Xu. Brushnet: A plug-and-play image inpainting model with decomposed dual-branch diffusion. InEuropean Conference on Computer Vision, pages 150–168. Springer, 2024. 11

    work page 2024

  16. [16]

    Turbofill: adapting few-step text- to-image model for fast image inpainting

    Liangbin Xie, Daniil Pakhomov, Zhonghao Wang, Zongze Wu, Ziyan Chen, Yuqian Zhou, Haitian Zheng, Zhifei Zhang, Zhe Lin, Jiantao Zhou, et al. Turbofill: adapting few-step text- to-image model for fast image inpainting. InProceedings of the Computer Vision and Pattern Recognition Conference, pages 7613–7622, 2025

    work page 2025

  17. [17]

    Repainter: Empowering e-commerce object removal via spatial-matting reinforcement learning.arXiv preprint arXiv:2510.07721, 2025

    Zipeng Guo, Lichen Ma, Xiaolong Fu, Gaojing Zhou, Lan Yang, Yuchen Zhou, Linkai Liu, Yu He, Ximan Liu, Shiping Dong, et al. Repainter: Empowering e-commerce object removal via spatial-matting reinforcement learning.arXiv preprint arXiv:2510.07721, 2025

    work page arXiv 2025

  18. [18]

    UM-Text: A Unified Multimodal Model for Image Understanding and Visual Text Editing

    Lichen Ma, Xiaolong Fu, Gaojing Zhou, Zipeng Guo, Ting Zhu, Yichun Liu, Yu Shi, Jason Li, and Junshi Huang. Um-text: A unified multimodal model for image understanding.arXiv preprint arXiv:2601.08321, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  19. [19]

    Replan: Reasoning-guided region planning for com- plex instruction-based image editing.arXiv preprint arXiv:2512.16864, 2025

    Tianyuan Qu, Lei Ke, Xiaohang Zhan, Longxiang Tang, Yuqi Liu, Bohao Peng, Bei Yu, Dong Yu, and Jiaya Jia. Replan: Reasoning-guided region planning for complex instruction-based image editing.arXiv preprint arXiv:2512.16864, 2025

    work page arXiv 2025

  20. [20]

    Got: Unleashing reasoning capability of multimodal large language model for visual generation and editing.arXiv preprint arXiv:2503.10639, 2025

    Rongyao Fang, Chengqi Duan, Kun Wang, Linjiang Huang, Hao Li, Shilin Yan, Hao Tian, Xingyu Zeng, Rui Zhao, Jifeng Dai, Xihui Liu, and Hongsheng Li. Got: Unleashing reasoning capability of multimodal large language model for visual generation and editing.arXiv preprint arXiv:2503.10639, 2025

    work page arXiv 2025

  21. [21]

    Mcie: Multimodal llm-driven complex instruction image editing with spatial guidance.arXiv preprint arXiv:2602.07993, 2026

    Xuehai Bai, Xiaoling Gu, Akide Liu, Hangjie Yuan, YiFan Zhang, and Jack Ma. Mcie: Multimodal llm-driven complex instruction image editing with spatial guidance.arXiv preprint arXiv:2602.07993, 2026

    work page arXiv 2026

  22. [22]

    arXiv preprint arXiv:2512.05965 (2025) 2, 3 PhyEditBench 17

    Hongyu Li, Manyuan Zhang, Dian Zheng, Ziyu Guo, Yimeng Jia, Kaituo Feng, Hao Yu, Yexin Liu, Yan Feng, Peng Pei, Xunliang Cai, Linjiang Huang, Hongsheng Li, and Si Liu. Editthinker: Unlocking iterative reasoning for any image editor.arXiv preprint arXiv:2512.05965, 2025

    work page arXiv 2025

  23. [23]

    arXiv preprint arXiv:2601.03467 (2026) 2

    Hengjia Li, Liming Jiang, Qing Yan, Yizhi Song, Hao Kang, Zichuan Liu, Xin Lu, Boxi Wu, and Deng Cai. Thinkrl-edit: Thinking in reinforcement learning for reasoning-centric image editing.arXiv preprint arXiv:2601.03467, 2026

    work page arXiv 2026

  24. [24]

    Posteromni: Generalized artistic poster creation via task distillation and unified reward feedback.arXiv preprint arXiv:2602.12127, 2026

    Sixiang Chen, Jianyu Lai, Jialin Gao, Hengyu Shi, Zhongying Liu, Tian Ye, Junfeng Luo, Xiaoming Wei, and Lei Zhu. Posteromni: Generalized artistic poster creation via task distillation and unified reward feedback.arXiv preprint arXiv:2602.12127, 2026

    work page arXiv 2026

  25. [25]

    Macro: Advancing multi-reference image generation with structured long-context data.arXiv preprint arXiv:2603.25319, 2026

    Zhekai Chen, Yuqing Wang, Manyuan Zhang, and Xihui Liu. Macro: Advancing multi-reference image generation with structured long-context data.arXiv preprint arXiv:2603.25319, 2026

    work page arXiv 2026

  26. [26]

    Prompt-to-Prompt Image Editing with Cross Attention Control

    Amir Hertz, Ron Mokady, Jay Tenenbaum, Kfir Aberman, Yael Pritch, and Daniel Cohen-Or. Prompt-to-prompt image editing with cross attention control.arXiv preprint arXiv:2208.01626, 2022

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  27. [27]

    Diffedit: Diffusion- based semantic image editing with mask guidance

    Guillaume Couairon, Jakob Verbeek, Holger Schwenk, and Matthieu Cord. Diffedit: Diffusion- based semantic image editing with mask guidance. InInternational Conference on Learning Representations, 2023

    work page 2023

  28. [28]

    Magicbrush: A manually annotated dataset for instruction-guided image editing

    Kai Zhang, Lingbo Mo, Wenhu Chen, Huan Sun, and Yu Su. Magicbrush: A manually annotated dataset for instruction-guided image editing. InAdvances in Neural Information Processing Systems, 2023

    work page 2023

  29. [29]

    In-context edit: Enabling instructional image editing with in-context generation in large scale diffusion transformer

    Zechuan Zhang, Ji Xie, Yu Lu, Zongxin Yang, and Yi Yang. In-context edit: Enabling instructional image editing with in-context generation in large scale diffusion transformer. In Advances in Neural Information Processing Systems, 2025

    work page 2025

  30. [30]

    Step1X-Edit: A Practical Framework for General Image Editing

    Shiyu Liu, Yucheng Han, Peng Xing, Fukun Yin, Rui Wang, Wei Cheng, Jiaqi Liao, Yingming Wang, Honghao Fu, Chunrui Han, Guopeng Li, Yuang Peng, Quan Sun, Jingwei Wu, Yan Cai, Zheng Ge, Ranchen Ming, Lei Xia, Xianfang Zeng, Yibo Zhu, Binxing Jiao, Xiangyu Zhang, Gang Yu, and Daxin Jiang. Step1x-edit: A practical framework for general image editing.arXiv pre...

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  31. [31]

    OmniGen2: Towards Instruction-Aligned Multimodal Generation

    Chenyuan Wu, Pengfei Zheng, Ruiran Yan, Shitao Xiao, Xin Luo, Yueze Wang, Wanli Li, Xiyan Jiang, Yexin Liu, Junjie Zhou, Ze Liu, Ziyi Xia, Chaofan Li, Haoge Deng, Jiahao Wang, Kun Luo, Bo Zhang, Defu Lian, Xinlong Wang, Zhongyuan Wang, Tiejun Huang, and Zheng Liu. Omnigen2: Towards instruction-aligned multimodal generation.arXiv preprint arXiv:2506.18871, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  32. [32]

    Emerging Properties in Unified Multimodal Pretraining

    Chaorui Deng, Deyao Zhu, Kunchang Li, Chenhui Gou, Feng Li, Zeyu Wang, Shu Zhong, Weihao Yu, Xiaonan Nie, Ziang Song, Guang Shi, and Haoqi Fan. Emerging properties in unified multimodal pretraining.arXiv preprint arXiv:2505.14683, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  33. [33]

    FLUX.2: Next generation image generation

    Black Forest Labs. FLUX.2: Next generation image generation. https://bfl.ai/models/ flux-2, 2025. Accessed: 2026-04-24

    work page 2025

  34. [34]

    Weedit: A dataset, benchmark and glyph-guided framework for text-centric image editing

    Hui Zhang, Juntao Liu, Zongkai Liu, Liqiang Niu, Fandong Meng, Zuxuan Wu, and Yu-Gang Jiang. Weedit: A dataset, benchmark and glyph-guided framework for text-centric image editing. arXiv preprint arXiv:2603.11593, 2026

    work page arXiv 2026

  35. [35]

    Grounding dino: Marrying dino with grounded pre-training for open-set object detection

    Shilong Liu, Zhaoyang Zeng, Tianhe Ren, Feng Li, Hao Zhang, Jie Yang, Qing Jiang, Chunyuan Li, Jianwei Yang, Hang Su, et al. Grounding dino: Marrying dino with grounded pre-training for open-set object detection. InEuropean Conference on Computer Vision, 2024

    work page 2024

  36. [36]

    Berg, Wan-Yen Lo, et al

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C. Berg, Wan-Yen Lo, et al. Segment anything. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 4015–4026, 2023

    work page 2023

  37. [37]

    Smartedit: Exploring complex instruction-based image editing with multimodal large language models

    Yuzhou Huang, Liangbin Xie, Xintao Wang, Ziyang Yuan, Xiaodong Cun, Yixiao Ge, Jiantao Zhou, Chao Dong, Rui Huang, Ruimao Zhang, and Ying Shan. Smartedit: Exploring complex instruction-based image editing with multimodal large language models. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024

    work page 2024

  38. [38]

    Fireedit: Fine-grained instruction-based image editing via region-aware vision language model

    Jun Zhou, Jiahao Li, Zunnan Xu, Hanhui Li, Yiji Cheng, Fa-Ting Hong, Qin Lin, Qinglin Lu, and Xiaodan Liang. Fireedit: Fine-grained instruction-based image editing via region-aware vision language model. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2025

    work page 2025

  39. [39]

    Beyond simple edits: X-planner for complex instruction-based image editing.arXiv preprint arXiv:2507.05259, 2025

    Chun-Hsiao Yeh, Yilin Wang, Nanxuan Zhao, Richard Zhang, Yuheng Li, Yi Ma, and Kr- ishna Kumar Singh. Beyond simple edits: X-planner for complex instruction-based image editing.arXiv preprint arXiv:2507.05259, 2025

    work page arXiv 2025

  40. [40]

    Xing, et al

    Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric P. Xing, et al. Judging llm-as-a-judge with mt-bench and chatbot arena. InAdvances in Neural Information Processing Systems, 2023

    work page 2023

  41. [41]

    Editreward: A human-aligned reward model for instruction-guided image editing, 2026

    Keming Wu, Sicong Jiang, Max Ku, Ping Nie, Minghao Liu, and Wenhu Chen. Editre- ward: A human-aligned reward model for instruction-guided image editing.arXiv preprint arXiv:2509.26346, 2025

    work page arXiv 2025

  42. [42]

    Trust your critic: Robust reward modeling and re- inforcement learning for faithful image editing and generation.arXiv preprint arXiv:2603.12247, 2026

    Xiangyu Zhao, Peiyuan Zhang, Junming Lin, Tianhao Liang, Yuchen Duan, Shengyuan Ding, Changyao Tian, Yuhang Zang, Junchi Yan, and Xue Yang. Trust your critic: Robust reward modeling and reinforcement learning for faithful image editing and generation.arXiv preprint arXiv:2603.12247, 2026

    work page arXiv 2026

  43. [43]

    Yolov10: Real-time end-to-end object detection.Advances in neural information processing systems, 37:107984–108011, 2024

    Ao Wang, Hui Chen, Lihao Liu, Kai Chen, Zijia Lin, Jungong Han, and Guiguang Ding. Yolov10: Real-time end-to-end object detection.Advances in neural information processing systems, 37:107984–108011, 2024

    work page 2024

  44. [44]

    SAM 2: Segment Anything in Images and Videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman Rädle, Chloe Rolland, Laura Gustafson, et al. Sam 2: Segment anything in images and videos.arXiv preprint arXiv:2408.00714, 2024. 13

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  45. [45]

    Pp-ocrv5: A specialized 5m-parameter model rivaling billion-parameter vision-language models on ocr tasks.arXiv preprint arXiv:2603.24373, 2026

    Cheng Cui, Yubo Zhang, Ting Sun, Xueqing Wang, Hongen Liu, Manhui Lin, Yue Zhang, Tingquan Gao, Changda Zhou, Jiaxuan Liu, et al. Pp-ocrv5: A specialized 5m-parameter model rivaling billion-parameter vision-language models on ocr tasks.arXiv preprint arXiv:2603.24373, 2026

    work page arXiv 2026

  46. [46]

    Qwen Team. Qwen3. 5-omni technical report.arXiv preprint arXiv:2604.15804, 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  47. [47]

    Qwen3-VL Technical Report

    Shuai Bai, Yuxuan Cai, Ruizhe Chen, Keqin Chen, Xionghui Chen, Zesen Cheng, Lianghao Deng, Wei Ding, Chang Gao, Chunjiang Ge, et al. Qwen3-vl technical report.arXiv preprint arXiv:2511.21631, 2025. 14 Appendix In this work, we propose GMO-E 2DIT, an agentic framework for grounded multi-operation e- commerce image editing. Due to space constraints in the m...

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  48. [48]

    edit_fidelity: How well does the edited result satisfy the requested edit inside the target regions? For insertion tasks, also judge whether the inserted content matches the patch image

    work page

  49. [49]

    In the precise protocol, examine each bounding box and reference region indices in the reasoning

    background_preservation: Outside the requested edit regions, is the edited image unchanged compared with the source image? Instructions.Read the benchmark instruction carefully. In the precise protocol, examine each bounding box and reference region indices in the reasoning. In the fuzzy protocol, identify the intended edit from the natural-language reque...

    work page

  50. [50]

    ungreenday,

    Unless otherwise specified, all experiments are conducted on the same constructed training dataset and evaluated on the held-out EComEditBench split. All experiments are conducted on 16 NVIDIA B200 GPUs. E More Experiment Results E.1 Multi-turn Editing To evaluate our multi-turn joint RL and reflection-driven mechanism, we conduct experiments on a test se...

    work page