PixelMan: Consistent Object Editing with Diffusion Models via Pixel Manipulation and Generation
Authors: Liyao Jiang, Negar Hassanpour, Mohammad Salameh, Mohammadreza Samadi, Jiao He, Fengyu Sun, Di Niu
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental evaluations based on benchmark datasets as well as extensive visual comparisons show that in as few as 16 inference steps, Pixel Man outperforms a range of state-of-the-art training-based and training-free methods (usually requiring 50 steps) on multiple consistent object editing tasks. Quantitative results on the COCOEE and Re S datasets as well as extensive visual comparisons suggest that Pixel Man achieves superior performance in consistency metrics for object, background, and semantic consistency between the source and edited image, while achieving higher or comparable performance in IQA metrics. |
| Researcher Affiliation | Collaboration | 1Department of Electrical and Computer Engineering, University of Alberta 2Huawei Technologies Canada 3Huawei Kirin Solution, China EMAIL EMAIL EMAIL |
| Pseudocode | Yes | Algorithm 1: Algorithm Overview of Pixel Man |
| Open Source Code | No | The paper provides a project page URL: "Project liyaojiang1998.github.io/projects/Pixel Man/". However, it does not explicitly state that source code is available at this link, nor does it provide a direct link to a code repository. The prompt specifies that project demonstration pages or high-level overviews are not sufficient unless they explicitly host source code. |
| Open Datasets | Yes | For our experiments, we adopted subsets of two challenging datasets, namely COCOEE (Lin et al. 2014; Yang et al. 2022) and Re S (Wang et al. 2024) (detailed in the Appendix). |
| Dataset Splits | No | The paper mentions using subsets of COCOEE and Re S datasets and states that details are in the Appendix, but it does not provide specific dataset split information (e.g., percentages, sample counts, or explicit standard splits) within the main text. |
| Hardware Specification | No | The paper does not provide specific hardware details such as GPU or CPU models, memory specifications, or types of computing infrastructure used for the experiments. |
| Software Dependencies | No | The paper mentions using models like SDv1.5 and SDv2-Inpainting Model, and the Any Door method, but it does not specify software libraries with version numbers (e.g., Python, PyTorch, TensorFlow versions) that constitute ancillary software dependencies for replication. |
| Experiment Setup | No | The paper mentions performing experiments in "as few as 16 inference steps" and discusses metrics for efficiency (inference steps, NFEs, latency), but it does not provide specific experimental setup details such as hyperparameters (e.g., learning rates, batch sizes, optimizer settings) or other training configurations. |