Object Placement for Anything

2025-04-16

AI Summary

Existing discriminative models for foreground object placement (position and scale) in background images exhibit weak generalization under few-shot settings. Method: This paper proposes a semi-supervised framework tailored to real-world scenarios, centered on a novel “plausibility-change knowledge transfer” mechanism: it leverages regularities in plausibility label shifts induced by fine-tuning foreground pose on labeled data to guide pseudo-label generation and model optimization on unlabeled data. The approach integrates discriminative plausibility modeling, pose-sensitivity modeling, and knowledge distillation. Contribution/Results: With only 10% labeled data, our method achieves state-of-the-art fully supervised performance; cross-domain generalization error is reduced by 32%, significantly enhancing robustness and practicality in few-shot regimes.

Technology Category

• Machine Learning: Semi-Supervised Learning
• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Learning to Search

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Object placement aims to determine the appropriate placement (emph{e.g.}, location and size) of a foreground object when placing it on the background image. Most previous works are limited by small-scale labeled dataset, which hinders the real-world application of object placement. In this work, we devise a semi-supervised framework which can exploit large-scale unlabeled dataset to promote the generalization ability of discriminative object placement models. The discriminative models predict the rationality label for each foreground placement given a foreground-background pair. To better leverage the labeled data, under the semi-supervised framework, we further propose to transfer the knowledge of rationality variation, emph{i.e.}, whether the change of foreground placement would result in the change of rationality label, from labeled data to unlabeled data. Extensive experiments demonstrate that our framework can effectively enhance the generalization ability of discriminative object placement models.

Problem

• Determining optimal placement of foreground objects on backgrounds
• Overcoming limitations of small labeled datasets for object placement
• Enhancing generalization using semi-supervised learning and rationality variation

Innovation

• Semi-supervised framework for object placement
• Exploits large-scale unlabeled dataset
• Transfers knowledge of rationality variation