Pixel-level Quality Assessment for Oriented Object Detection

2025-11-11

AI Summary

Existing rotated object detectors rely on box-level IoU prediction to assess localization quality; however, such metrics exhibit structural coupling with ground-truth localization accuracy, leading to systematic overestimation of low-quality predictions. To address this, we propose Pixel-level Quality Assessment (PQA), a novel framework that abandons box-level IoU regression and instead models the spatial consistency between predicted and ground-truth rotated bounding boxes in pixel space—enabling decoupled and robust localization quality estimation. PQA employs a spatial consistency aggregation mechanism to produce a unified quality score and can be seamlessly integrated as a plug-and-play module into mainstream rotated detectors. Extensive experiments on HRSC2016 and DOTA demonstrate substantial improvements: PQA boosts Rotated RetinaNet’s AP$_{50:95}$ by 5.96% and reduces standard deviation (STD) by 2.32%, validating both the effectiveness and generalizability of pixel-level quality modeling.

Technology Category

• Knowledge Representation and Reasoning: Qualitative Reasoning
• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Low Level & Physics-based Vision

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization

Abstract

Modern oriented object detectors typically predict a set of bounding boxes and select the top-ranked ones based on estimated localization quality. Achieving high detection performance requires that the estimated quality closely aligns with the actual localization accuracy. To this end, existing approaches predict the Intersection over Union (IoU) between the predicted and ground-truth (GT) boxes as a proxy for localization quality. However, box-level IoU prediction suffers from a structural coupling issue: since the predicted box is derived from the detector's internal estimation of the GT box, the predicted IoU--based on their similarity--can be overestimated for poorly localized boxes. To overcome this limitation, we propose a novel Pixel-level Quality Assessment (PQA) framework, which replaces box-level IoU prediction with the integration of pixel-level spatial consistency. PQA measures the alignment between each pixel's relative position to the predicted box and its corresponding position to the GT box. By operating at the pixel level, PQA avoids directly comparing the predicted box with the estimated GT box, thereby eliminating the inherent similarity bias in box-level IoU prediction. Furthermore, we introduce a new integration metric that aggregates pixel-level spatial consistency into a unified quality score, yielding a more accurate approximation of the actual localization quality. Extensive experiments on HRSC2016 and DOTA demonstrate that PQA can be seamlessly integrated into various oriented object detectors, consistently improving performance (e.g., +5.96% AP$_{50:95}$ on Rotated RetinaNet and +2.32% on STD).

Problem

• Overcoming structural coupling in box-level IoU prediction for object detection
• Replacing box-level quality assessment with pixel-level spatial consistency
• Improving localization quality estimation for oriented object detectors

Innovation

• Pixel-level spatial consistency replaces box-level IoU prediction
• Integration metric aggregates pixel alignment into quality score
• Framework eliminates similarity bias in localization quality assessment