Abstract:Deep learning has proven an effective tool for 3D point cloud processing. Currently, most deep set architectures are developed for sparse inputs (typically with a few thousand points), which are unable to provide sufficient structural statistics and semantic cues due to low resolutions. Since these architectures suffer from unacceptable computational and memory costs when consuming dense inputs, there is a pressing need in real-world applications to handle large-scale 3D point clouds. To bridge this gap, this paper presents a novel unsupervised neural architecture called RegGeoNet to parameterize an unstructured point set into a completely regular image structure dubbed as deep geometry image (DeepGI), such that spatial coordinates of unordered points are recorded in three-channel grid pixels. Intuitively, our goal is to embed irregular 3D surface points onto uniform 2D lattice grids, while trying to preserve local neighborhood consistency. Functionally, DeepGI serves as a generic representation modality for raw point cloud data and can be conveniently integrated into mature image processing pipelines. Driven by its unique structural characteristics, we are motivated to customize a set of efficient feature extractors that directly operate on DeepGIs for achieving a rich variety of downstream tasks. To demonstrate the potential and universality of our proposed learning paradigms built upon DeepGIs for large-scale point cloud processing, we conduct extensive experiments on various downstream tasks, including shape classification, object part segmentation, scene semantic segmentation, normal estimation, and geometry compression, where our frameworks achieve highly competitive performance, compared with state-of-the-art methods. The source code will be publicly available at https://github.com/keeganhk/RegGeoNet.

Image Retrieval by Geological Proximity Using Deep Neural Network.

Superpixel-Based Long-Range Dependent Network for High-Resolution Remote-Sensing Image Classification

Wide-Area Image Geolocalization with Aerial Reference Imagery

Deep Neural Network Retrieval.

Bag-Of-Words Based Deep Neural Network For Image Retrieval

Multi-View 3d Object Retrieval with Deep Embedding Network

Retrieval Across Optical and SAR Images with Deep Neural Network.

Content-Based Image Retrieval for Construction Site Images: Leveraging Deep Learning–Based Object Detection

RegGeoNet: Learning Regular Representations for Large-Scale 3D Point Clouds

Enhancing Remote Sensing Image Retrieval: A Hierarchical Approach Integrating Visual and Semantic Similarities

Deep Correlated Joint Network for 2-D Image-Based 3-D Model Retrieval

One net to rule them all: efficient recognition and retrieval of POI from geo-tagged photos

Codedretrieval: Joint Image Compression and Retrieval with Neural Networks.

Img2Loc: Revisiting Image Geolocalization using Multi-modality Foundation Models and Image-based Retrieval-Augmented Generation

Deep Learning Approach for SAR Image Retrieval for Reliable Positioning in GPS-Challenged Environments

Beyond Cross-view Image Retrieval: Highly Accurate Vehicle Localization Using Satellite Image

IML-Net: A Framework for Cross-View Geo-Localization with Multi-Domain Remote Sensing Data

Differential Geometric Retrieval of Deep Features

Pixel-Wise Classification Method For High Resolution Remote Sensing Imagery Using Deep Neural Networks

Enhancing Remote Sensing Image Retrieval with Triplet Deep Metric Learning Network

An Efficient Approach for Geo-Multimedia Cross-Modal Retrieval