Prompt-Driven Simulation with Feature Perturbation for Cross-Domain Few-Shot Object Detection

2026-08-02

AI Summary

This work addresses the performance degradation in cross-domain few-shot object detection (CD-FSOD) caused by severe domain shift and scarce annotations in the target domain. To tackle these challenges, the authors propose the PSP-FSOD framework, which innovatively integrates prompt-driven foreground-background joint modeling with a grounding-aware image generation strategy to synthesize semantically consistent yet domain-diverse training images using large vision-language models. Additionally, the framework incorporates a multi-scale feature perturbation module coupled with a distribution alignment mechanism to enhance domain-invariant representation learning. Extensive experiments demonstrate that PSP-FSOD significantly outperforms existing data augmentation approaches across multiple CD-FSOD benchmarks, effectively improving model generalization and robustness under limited supervision and substantial domain discrepancy.

Abstract

Data augmentation, which simulates diverse visual variations to expand the source distribution and induce synthetic domain shifts, is a simple yet effective strategy for mitigating severe domain shifts and limited labeled target data in cross-domain few-shot object detection (CD-FSOD). Existing approaches rely on conventional data augmentation, such as Color-Jitter, Mosaic, and background-centric adaptation (e.g., Domain-RAG), which are limited in modeling complex domain shifts and often lead to suboptimal performance. In this paper, we propose PSP-FSOD, a principled framework that integrates prompt-driven domain simulation with feature perturbation regularization to improve generalization in CD-FSOD. To enable controllable domain synthesis, we design a prompt-driven strategy that leverages the visual grounding capability of large VLMs to jointly model foreground and background variations, generating semantically consistent yet domain-diverse training samples. Moreover, we adopt a grounding-aware generation scheme that guides object placement and alleviates semantic-spatial misalignment, thereby improving foreground adaptation. To ensure training stability and robustness, we further introduce a noise-induced feature perturbation mechanism that injects Gaussian noise into multi-scale intermediate features with distribution correction, encouraging consistent predictions under perturbations and reducing reliance on domain-specific cues. Extensive experiments demonstrate that PSP-FSOD produces high-quality domain-diverse supervision and learns domain-invariant representations, consistently improving performance across CD-FSOD benchmarks.

Problem

• cross-domain few-shot object detection
• domain shift
• data augmentation
• domain generalization
• limited labeled data

Innovation

• prompt-driven simulation
• feature perturbation
• visual grounding
• cross-domain few-shot object detection
• domain-invariant representation