P3S-Diffusion:A Selective Subject-driven Generation Framework via Point Supervision

TL;DR

P3S-Diffusion is a new framework for subject-driven image generation that uses minimal point supervision to accurately select and generate specific subjects, improving feature preservation and reducing reliance on costly segmentation methods.

Contribution

The paper introduces P3S-Diffusion, a novel architecture that utilizes point supervision for selective subject-driven generation, eliminating the need for additional segmentation models during fine-tuning.

Findings

Effective in preserving fine subject features.

Capable of generating high-quality subject-specific images.

Reduces costs by avoiding pixel masks and segmentation models.

Abstract

Recent research in subject-driven generation increasingly emphasizes the importance of selective subject features. Nevertheless, accurately selecting the content in a given reference image still poses challenges, especially when selecting the similar subjects in an image (e.g., two different dogs). Some methods attempt to use text prompts or pixel masks to isolate specific elements. However, text prompts often fall short in precisely describing specific content, and pixel masks are often expensive. To address this, we introduce P3S-Diffusion, a novel architecture designed for context-selected subject-driven generation via point supervision. P3S-Diffusion leverages minimal cost label (e.g., points) to generate subject-driven images. During fine-tuning, it can generate an expanded base mask from these points, obviating the need for additional segmentation models. The mask is employed for inpainting and aligning with subject representation. The P3S-Diffusion preserves fine features of the subjects through Multi-layers Condition Injection. Enhanced by the Attention Consistency Loss for improved training, extensive experiments demonstrate its excellent feature preservation and image generation capabilities.