Pattern-Enhanced RT-DETR for Multi-Class Battery Detection

2026-05-13

AI Summary

This work addresses the performance degradation in detecting data-scarce battery categories—such as bicycle batteries—within multi-class battery detection tasks by establishing the first comprehensive battery detection benchmark and proposing PaQ-RT-DETR. The method integrates a pattern-aware dynamic query generation mechanism into the RT-DETR framework, effectively mitigating query activation imbalance while introducing negligible computational overhead. Experimental results demonstrate that PaQ-RT-DETR-X achieves an mAP@50 of 0.782 across six battery categories, representing a 2.8% improvement over RT-DETR-X, with consistent performance gains observed across all classes, including those with limited training data.

Abstract

Accurate and efficient battery detection is increasingly important for applications in electronic waste recycling, industrial quality control, and automated sorting systems. In this paper, we present both a comprehensive benchmark and a novel method for multi-class battery detection. We systematically compare three CNN-based detectors (YOLOv8n, YOLOv8s, YOLO11n) and two transformer-based detectors (RT-DETR-L, RT-DETR-X) on a publicly available dataset of approximately 8,591 annotated images under identical experimental conditions, and further propose PaQ-RT-DETR, which introduces pattern-based dynamic query generation into RT-DETR to alleviate query activation imbalance with negligible computational overhead. Among baselines, YOLO11n achieves the best CNN-based accuracy (mAP@50: 0.779) at only 2.6M parameters, while YOLOv8n delivers the fastest inference at ~1,667 FPS. PaQ-RT-DETR-X achieves the highest overall mAP@50 of 0.782, surpassing RT-DETR-X by +2.8% with consistent per-class gains across all six battery categories including the data-scarce Bike Battery class. Our findings provide practical guidance for selecting object detection models in battery-related industrial applications.

Problem

• battery detection
• multi-class detection
• object detection
• industrial applications
• electronic waste recycling

Innovation

• pattern-based dynamic query
• RT-DETR
• multi-class battery detection
• query activation imbalance
• object detection