Deep learning model advances utility pole and sign detection

Researchers have developed a deep learning framework for detecting, segmenting, and estimating the lean angle of utility poles, as well as classifying attached warning signs. The system, based on a modified Detection Transformer (DETR) model, was trained on a custom dataset of 4,570 annotated images from Google Street View. This approach surpasses standard object detectors like RetinaNet and YOLOv3-Tiny, achieving high mean average precision for both pole and sign detection. The framework also enables mask generation for accurate pole lean angle estimation, with a low mean absolute error on the test set. AI

IMPACT This research could improve infrastructure inspection efficiency and safety through automated analysis of ground-level imagery.

RANK_REASON Academic paper detailing a new deep learning model and dataset for a specific computer vision task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• Detection Transformer
• Faster R-CNN
• Google Street View
• Hugging Face
• RetinaNet
• YOLOv3-Tiny

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →