HiFi-Inpaint: Towards High-Fidelity Reference-Based Inpainting for Generating Detail-Preserving Human-Product Images

2026-03-02

AI Summary

This work addresses the challenge of faithfully preserving fine-grained product details in reference-guided image inpainting for human–product imagery, a task hindered by insufficient training data, limited model sensitivity to high-frequency details, and coarse supervision signals. To overcome these limitations, we propose HiFi-Inpaint, a novel framework that introduces a shared augmented attention mechanism and a detail-aware loss function. Additionally, we construct HP-Image-40K, a large-scale, high-quality dataset specifically curated for this task. By integrating reference-guided inpainting architecture, explicit high-frequency detail supervision, and an automated synthetic data filtering strategy, HiFi-Inpaint significantly outperforms existing methods, achieving markedly more realistic and intricate reconstruction of product details in human–product image generation.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Human-product images, which showcase the integration of humans and products, play a vital role in advertising, e-commerce, and digital marketing. The essential challenge of generating such images lies in ensuring the high-fidelity preservation of product details. Among existing paradigms, reference-based inpainting offers a targeted solution by leveraging product reference images to guide the inpainting process. However, limitations remain in three key aspects: the lack of diverse large-scale training data, the struggle of current models to focus on product detail preservation, and the inability of coarse supervision for achieving precise guidance. To address these issues, we propose HiFi-Inpaint, a novel high-fidelity reference-based inpainting framework tailored for generating human-product images. HiFi-Inpaint introduces Shared Enhancement Attention (SEA) to refine fine-grained product features and Detail-Aware Loss (DAL) to enforce precise pixel-level supervision using high-frequency maps. Additionally, we construct a new dataset, HP-Image-40K, with samples curated from self-synthesis data and processed with automatic filtering. Experimental results show that HiFi-Inpaint achieves state-of-the-art performance, delivering detail-preserving human-product images.

Problem

• reference-based inpainting
• human-product images
• detail preservation
• high-fidelity generation

Innovation

• reference-based inpainting
• Shared Enhancement Attention
• Detail-Aware Loss
• high-fidelity image generation
• human-product image synthesis