GridNet-HD: A High-Resolution Multi-Modal Dataset for LiDAR-Image Fusion on Power Line Infrastructure

2026-01-19

AI Summary

This work addresses the absence of publicly available multimodal datasets that simultaneously incorporate high-density LiDAR, high-resolution oblique imagery, and fine-grained 3D semantic annotations for power line infrastructure. To bridge this gap, we introduce GridNet-HD, a novel benchmark dataset comprising 250 million LiDAR points and 7,694 oblique images, annotated with 11 semantic classes in 3D—the first of its kind to be released. Leveraging this dataset, we develop both single-modality and multimodal fusion baselines for 3D semantic segmentation, demonstrating the complementary nature of geometric and appearance cues. Experimental results show that the multimodal model achieves a 5.55% improvement in mean Intersection over Union (mIoU) over the best single-modality baseline, significantly enhancing 3D semantic understanding of power line assets.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

This paper presents GridNet-HD, a multi-modal dataset for 3D semantic segmentation of overhead electrical infrastructures, pairing high-density LiDAR with high-resolution oblique imagery. The dataset comprises 7,694 images and 2.5 billion points annotated into 11 classes, with predefined splits and mIoU metrics. Unimodal (LiDAR-only, image-only) and multi-modal fusion baselines are provided. On GridNet-HD, fusion models outperform the best unimodal baseline by +5.55 mIoU, highlighting the complementarity of geometry and appearance. As reviewed in Sec. 2, no public dataset jointly provides high-density LiDAR and high-resolution oblique imagery with 3D semantic labels for power-line assets. Dataset, baselines, and codes are available: https://huggingface.co/collections/heig-vd-geo/gridnet-hd.

Problem

• LiDAR-image fusion
• power line infrastructure
• 3D semantic segmentation
• multi-modal dataset
• high-resolution imagery

Innovation

• multi-modal fusion
• high-density LiDAR
• high-resolution oblique imagery
• 3D semantic segmentation
• power line infrastructure