A5029685813- 3D Shape Modeling and Analysis
- Advanced Image and Video Retrieval Techniques
- Image Retrieval and Classification Techniques
- Image Processing and 3D Reconstruction
- Robotics and Sensor-Based Localization
- Landslides and related hazards
- 3D Surveying and Cultural Heritage
- Advanced Vision and Imaging
- Computer Graphics and Visualization Techniques
- Earthquake and Disaster Impact Studies
- Tree Root and Stability Studies
- Soil and Unsaturated Flow
- Cryospheric studies and observations
- Soil erosion and sediment transport
- Energy, Environment, Agriculture Analysis
- Medical Imaging and Analysis
- Video Surveillance and Tracking Methods
- Urban and spatial planning
- Medical Image Segmentation Techniques
- Fire effects on ecosystems
- Rangeland Management and Livestock Ecology
- Video Analysis and Summarization
- Seismology and Earthquake Studies
- Climate change and permafrost
- Meteorological Phenomena and Simulations
University of Yamanashi
2013-2024
Takeda (Japan)
2009-2024
Shibuya University Network
2023
Shimonoseki City Hospital
2016
Science Applications International Corporation (United States)
2011
Chiba University
1992-2005
Jikei University Kashiwa hospital
2000
Jikei University School of Medicine
2000
Mie University
1982
In this paper, we describe a shape-based 3D model retrieval method based on multi-scale local visual features. The features are extracted from 2D range images of the viewed uniformly sampled locations view sphere. is appearance-based, and accepts all models that can be rendered as image. For each image, set computed by using Scale Invariant Feature Transform [22] algorithm. To reduce cost distance computation feature storage, describing integrated into histogram Bag-Of-Features approach. Our...
Our previous shape-based 3D model retrieval algorithm compares shapes by using thousands of local visual features per model. A is rendered into a set depth images, and from each image, are extracted the Scale Invariant Feature Transform (SIFT) Lowe. To efficiently compare among large sets features, employs bag-of-features approach to integrate feature vector The outperformed other methods for dataset containing highly articulated yet geometrically simple models. For diverse detailed models,...
Non-rigid 3D shape retrieval has become a research hotpot in communities of computer graphics, vision, pattern recognition, etc. In this paper, we present the results SHREC'15 Track: Shape Retrieval. The aim track is to provide fair and effective platform evaluate compare performance current non-rigid methods developed by different groups around world. database utilized consists 1200 watertight triangle meshes which are equally classified into 50 categories. All models same category...
Non-rigid shape matching is one of the most challenging fields in content-based 3D object retrieval. The aim Shape Retrieval Contest 2010 (SHREC'10) track on non-rigid retrieval to evaluate and compare effectiveness different methods run a benchmark consisting 200 watertight triangular meshes. Three groups with six have participated this performance was evaluated using commonly-used metrics.
This paper describes a 3D shape model retrieval method that accepts, as query, mesh obtained by range scan from viewpoint. The proposed visually compares single depth map of the query with maps rendered multiple viewpoints. Comparison employs bag-of local visual features extracted using modified version Lowe's Scale-Invariant Feature Transform (SIFT). is capable retrieving models having diverse representations and robust against articulation global deformation shapes thanks to location-free...
Generic 3D shape retrieval is a fundamental research area in the field of content-based model retrieval. The aim this track to measure and compare performance generic methods implemented by different participants over world. based on new benchmark, which contains 1200 triangle meshes that are equally classified into 60 categories. In track, 16 runs have been submitted 5 groups their accuracies were evaluated using 7 commonly used metrics.
Sketch-based 3D model retrieval algorithms compare a query, line drawing sketch, and models for similarity by rendering the into drawing-like images. Still, accuracies of previous remained low, as sets features, one sketches other rendered images models, are quite different, they said to lie in different domains. A approach used semantic labels establish correspondence between features across inter-domain gap. This approach, however, is prone over learning if dataset difficult learn, i.e.,...
Large scale sketch-based 3D shape retrieval has received more and attentions in the community of content-based object retrieval. The objective this track is to evaluate performance different model algorithms using a large hand-drawn sketch query dataset on comprehensive dataset. benchmark contains 12,680 sketches 8,987 models, divided into 171 distinct classes. In track, 12 runs were submitted by 4 groups their was evaluated 7 commonly used metrics. We hope that benchmark, comparative...
Parameter-efficient fine-tuning (PEFT) of pre-trained 3D point cloud Transformers has emerged as a promising technique for analysis. While existing PEFT methods attempt to minimize the number tunable parameters, they still suffer from high temporal and spatial computational costs during fine-tuning. This paper proposes novel algorithm Transformers, called Side Token Adaptation on neighborhood Graph (STAG), achieve superior efficiency. STAG employs graph convolutional side network that...
This paper proposes a 3D model retrieval algorithm that employs an unsupervised distance metric learning with combination of appearance-based features; two sets local visual features and set global features. These are extracted from range images rendered multiple viewpoints about the to be compared. The bag-of-features histograms Scale Invariant Feature Transform (SIFT) by Lowe [7] sampled at either salient or dense random points. feature is also SIFT image center. proposed method then uses...
Sketch-based 3D shape retrieval has become an important research topic in content-based object retrieval. The aim of this track is to measure and compare the performance sketch-based methods implemented by different participants over world. based on a new benchmark, which contains two types sketch queries versions target models. In track, 7 runs have been submitted 5 groups their accuracies were evaluated using commonly used metrics. We hope that its corresponding evaluation code,...
With the advent of commodity 3D capturing devices and better modeling tools, shape content is becoming increasingly prevalent. Therefore, need for retrieval algorithms to handle large-scale repositories more important. This track provides a benchmark evaluate based on ShapeNet dataset. It continuation SHREC 2016 challenge with goal measuring progress recent developments in deep learning methods retrieval. We use Core55, which than 50 thousands models over 55 common categories total training...
Aggregating a set of local features has become one the most common approaches for representing multi-media data such as 2D image and 3D model. The success Bag-of-Features (BF) aggregation [2] prompted several extensions to BF, that are, VLAD [12], Fisher Vector (FV) coding [22] Super (SV) [34]. They all learn small number codewords, or representative features, by clustering large features. extracted from media (e.g., an image) is encoded considering distribution around codewords; BF uses...
Various algorithms for shape-based retrieval of non-rigid 3D models, with invariance to articulation and/or global deformation, have been developed. A majority these assumes that models mathematically well-defined representations, e.g., closed, manifold mesh. These are thus not applicable other types shape example, those defined as polygon soup. This paper proposes a model algorithm accepts diverse representations and is able compare models. The employs set hundreds thousands 3D,...