Preserving Product Fidelity in Large Scale Image Recontextualization with Diffusion Models
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929754111737856 |
|---|---|
| author | Malhi, Ishaan Dutta, Praneet Talius, Ellie Ma, Sally Driscoll, Brendan Holden, Krista Pruthi, Garima Narayanaswamy, Arunachalam |
| author_facet | Malhi, Ishaan Dutta, Praneet Talius, Ellie Ma, Sally Driscoll, Brendan Holden, Krista Pruthi, Garima Narayanaswamy, Arunachalam |
| contents | We present a framework for high-fidelity product image recontextualization using text-to-image diffusion models and a novel data augmentation pipeline. This pipeline leverages image-to-video diffusion, in/outpainting & negatives to create synthetic training data, addressing limitations of real-world data collection for this task. Our method improves the quality and diversity of generated images by disentangling product representations and enhancing the model's understanding of product characteristics. Evaluation on the ABO dataset and a private product dataset, using automated metrics and human assessment, demonstrates the effectiveness of our framework in generating realistic and compelling product visualizations, with implications for applications such as e-commerce and virtual product showcasing. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_08729 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Preserving Product Fidelity in Large Scale Image Recontextualization with Diffusion Models Malhi, Ishaan Dutta, Praneet Talius, Ellie Ma, Sally Driscoll, Brendan Holden, Krista Pruthi, Garima Narayanaswamy, Arunachalam Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning We present a framework for high-fidelity product image recontextualization using text-to-image diffusion models and a novel data augmentation pipeline. This pipeline leverages image-to-video diffusion, in/outpainting & negatives to create synthetic training data, addressing limitations of real-world data collection for this task. Our method improves the quality and diversity of generated images by disentangling product representations and enhancing the model's understanding of product characteristics. Evaluation on the ABO dataset and a private product dataset, using automated metrics and human assessment, demonstrates the effectiveness of our framework in generating realistic and compelling product visualizations, with implications for applications such as e-commerce and virtual product showcasing. |
| title | Preserving Product Fidelity in Large Scale Image Recontextualization with Diffusion Models |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2503.08729 |