State-of-the-art approaches for crowd counting resort to deepneural networks predict density maps. However, people in congested scenes remains a challenging task because the presence of drastic scale variation, inconsistency, and complex background can seriously degrade their accuracy. To battle ingrained issue accuracy degradation, this paper, we propose novel powerful network called Scale Tree Network (STNet) accurate counting. STNet consists two key components: Scale-Tree Diversity...

Point cloud processing is rapidly expanding the applicable scenarios in industry. The surface normal a fundamental feature for various point tasks. Recently, deep supervised estimators outperform traditional estimation methods by adapting to dataset statistics. However, existing mainly adopt hand-crafted features or complicated networks borrowed from other tasks and make less effort design network models specifically problem. Instead of regressing vector directly, we propose simple estimate...

Point cloud segmentation is one of the most important tasks in LiDAR remote sensing with widespread scientific, industrial, and commercial applications. The research thereof has resulted many breakthroughs 3D object scene understanding. Existing methods typically utilize hierarchical architectures for feature representation. However, commonly used sampling grouping networks are not only time-consuming but also limited to point-wise coordinates, ignoring local semantic homogeneity point...

In recent years, multi-sensor fusion technology has made enormous progress in 3D reconstruction, surveying and mapping, autonomous driving, other related fields, extrinsic calibration is a necessary condition for applications. This paper proposes LIDAR-to-camera automatic framework based on graph optimization. The system can automatically identify the position of pattern build set virtual feature point clouds, simultaneously complete LIDAR multiple cameras. To test this framework, formed...

Abstract In this paper, we propose a multi-level critical point aggregation architecture for 3D cloud normal estimation. It efficiently focuses on locally important points during feature extraction by employing our Local Feature Aggregation (LFA) and Global Refinement (GFR) modules. These modules can accurately identify surface-fitting across local global levels. Specifically, the proposed LFA module aims to capture geometric information from nearby with strong correlation in low-level...

Reinforced concrete structures play a pivotal role in island and reef engineering projects. Given the resource constraints typical of regions, substituting traditional manufactured sand aggregate with limestone not only reduces reliance on river but also addresses issue disposing waste slag generated during excavation. However, performance characteristics concrete, particularly its bond strength reinforcing steel, warrant further investigation. This is true for bond–slip behavior...

In recent years, deep learning-based point cloud normal estimation has made great progress. However, existing methods mainly rely on the PCPNet dataset, leading to overfitting. addition, correlation between clouds with different noise scales remains unexplored, resulting in poor performance cross-domain scenarios. this paper, we explore consistency of intrinsic features learned from clean and noisy using an Asymmetric Siamese Network architecture. By applying reasonable constraints extracted...

Directivity in magnetostrictive fiber-optic interferometric transducers was analyzed. Comparison carried out theoretically and experimentally on cylindrical racetrack transducers, expressions were presented. The results show that the transducer has better performance both sensitivity directivity for it a length of strip parallel to measured field

With the development of release and retrieval technology for unmanned air vehicles (UAVs), aerial cluster platform composed carrier aircraft multiple rotor UAVs can be used as a new approach mobile emergency broadcasting operations. It indicates that we resolve paradoxical relationship between limited endurance large scale mission area. This paper proposes region scanning strategy based on UAV platform. The system achieve all-region in disaster ares, well audio video collection assessment rescue.

Canonical extrinsic representations for non-rigid shapes with different poses are preferable in many computer graphics applications, such as shape correspondence and retrieval. The main reason this is that they give a pose invariant signature those jobs, which significantly decreases the difficulty caused by various poses. Existing methods based on multidimentional scaling (MDS) always result significant geometric distortions. In paper, we present novel unfolding algorithm, deforms any given...

With the burst development of neural networks in recent years, task normal estimation has once again become a concern. By introducing to classic methods based on problem-specific knowledge, adaptability algorithm noise and scale been greatly improved. However, compatibility between traditional not considered. Similar principle Occam's razor, that is, simpler is better. We observe more simplified process surface fitting can significantly improve accuracy estimation. In this paper, two...