Depth as Prior Knowledge for Object Detection

2026-02-05

AI Summary

This work addresses the challenge of accurately detecting small-scale and distant objects—a critical limitation in safety-sensitive applications due to scale variation, low resolution, and background clutter. The authors propose DepthPrior, a novel framework that leverages depth information as a supervisory prior rather than for feature fusion, enabling plug-and-play performance gains without altering the detector architecture. By introducing depth-guided loss weighting (DLW), loss stratification (DLS), and confidence threshold adjustment (DCT), the method significantly enhances detection robustness. Evaluated on four benchmarks including KITTI and MS COCO, DepthPrior achieves up to 9% and 7% improvements in mAP_S and mAR_S for small objects, respectively, with a true-to-false detection recovery ratio of 95:1—all without additional sensors or computational overhead.

Technology Category

• Computer Vision: Object Detection & Categorization
• Search and Optimization: Learning to Search
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Detecting small and distant objects remains challenging for object detectors due to scale variation, low resolution, and background clutter. Safety-critical applications require reliable detection of these objects for safe planning. Depth information can improve detection, but existing approaches require complex, model-specific architectural modifications. We provide a theoretical analysis followed by an empirical investigation of the depth-detection relationship. Together, they explain how depth causes systematic performance degradation and why depth-informed supervision mitigates it. We introduce DepthPrior, a framework that uses depth as prior knowledge rather than as a fused feature, providing comparable benefits without modifying detector architectures. DepthPrior consists of Depth-Based Loss Weighting (DLW) and Depth-Based Loss Stratification (DLS) during training, and Depth-Aware Confidence Thresholding (DCT) during inference. The only overhead is the initial cost of depth estimation. Experiments across four benchmarks (KITTI, MS COCO, VisDrone, SUN RGB-D) and two detectors (YOLOv11, EfficientDet) demonstrate the effectiveness of DepthPrior, achieving up to +9% mAP$_S$ and +7% mAR$_S$ for small objects, with inference recovery rates as high as 95:1 (true vs. false detections). DepthPrior offers these benefits without additional sensors, architectural changes, or performance costs. Code is available at https://github.com/mos-ks/DepthPrior.

Problem

• object detection
• small objects
• depth information
• scale variation
• background clutter

Innovation

• DepthPrior
• object detection
• depth as prior
• small object detection
• loss weighting