检索结果-内蒙古大学图书馆

Activation functions in deep learning: A comprehensive survey and benchmark

NEUROCOMPUTING 2022年 503卷 92-108页

作者： Dubey, Shiv Ram Singh, Satish Kumar Chaudhuri, Bidyut Baran Indian Inst Informat Technol Comp Vis & Biometr Lab Allahabad India Techno India Univ Kolkata India Indian Stat Inst Kolkata India Indian Inst Informat Technol Allahabad Allahabad India

Neural networks have shown tremendous growth in recent years to solve numerous problems. Various types of neural networks have been introduced to deal with different types of problems. However, the main goal of any neural network is to transform the non-linearly separable input data into more linearly separable abstract features using a hierarchy of layers. These layers are combinations of linear and non-linear functions. The most popular and common non-linearity layers are activation functions (AFs), such as Logistic Sigmoid, Tanh, ReLU, ELU, Swish and Mish. In this paper, a comprehensive overview and sur-vey is presented for AFs in neural networks for deep learning. Different classes of AFs such as Logistic Sigmoid and Tanh based, ReLU based, ELU based, and Learning based are covered. Several characteristics of AFs such as output range, monotonicity, and smoothness are also pointed out. A performance compar-ison is also performed among 18 state-of-the-art AFs with different networks on different types of data. The insights of AFs are presented to benefit the researchers for doing further research and practitioners to select among different choices. The code used for experimental comparison is released at: https://github.-com/shivram1987/ActivationFunctions.(c) 2022 Elsevier B.V. All rights reserved.

关键词： Activation Functions Neural networks Convolutional neural networks Deep learning Overview Recurrent Neural Networks

来源：评论

学校读者我要写书评

暂无评论

vision Transformer Hashing for image Retrieval

Vision Transformer Hashing for Image Retrieval

引用

2022 IEEE International conference on Multimedia and Expo, ICME 2022

作者： Dubey, Shiv Ram Singh, Satish Kumar Chu, Wei-Ta Indian Institute of Information Technology Computer Vision and Biometrics Lab Allahabad India National Cheng Kung University Dept. of Computer Science and Information Engineering Taiwan

ISBN: (数字)9781665485630

ISBN: (纸本)9781665485630

Recently, Transformer has emerged as a new architecture in deep learning by utilizing self-attention without convolution. Transformer is also extended to vision Transformer (ViT) for the visual recognition with a promising performance on imageNet. In this paper, we propose a vision Transformer Hashing (VTS) for image retrieval. We utilize the pre-trained ViT on imageNet as the backbone network and add the hashing head. The proposed VTS model is fine tuned for hashing under six different image retrieval frameworks with their objective functions. We perform the extensive experiments on CIFAR10, imageNet, NUS-Wide, and COCO datasets. The proposed VTS based image retrieval outperforms the recent state-of-the-art hashing techniques with a significant margin. We also find the proposed VTS model as the backbone network is better than the existing networks, such as AlexNet and ResNet. The code is released at https://***/shivram1987/visionTransformerHashing. © 2022 IEEE.

关键词： image retrieval

来源：评论

学校读者我要写书评

暂无评论

Aggregated Co-attention based Visual Question Answering 23

Aggregated Co-attention based Visual Question Answering

引用

proceedings of the Fourteenth indian conference on computer vision, graphics and image processing

作者： Aakansha Mishra Ashish Anand Prithwijit Guha IIT Guwahati India Indian Institute of Technology Guwahati IN IIT Guwahati IN

ISBN: (纸本)9798400716256

Recent developments in the field of Visual Question Answering (VQA) have witnessed promising improvements in performance through contributions in attention based networks. Most such approaches have focused on unidirectional attention that leverage over attention from textual domain (question) on visual space. This work proposes a multistage co-attention framework. Here, co-attention framework performs both image and text attention. The co-attention mechanism is repeated in multiple stages. Attention on different stages may capture some significant and distinct features for learning better contextual information. Thus, aggregation of attention is performed to preserve the information from different stages. The proposed architecture with multiple stage network could suffer from vanishing or exploding gradients. To prevent this, loss at each of the different stages is computed. Extensive experiments and analysis are performed for validating the effects of aggregated attention and stage-wise loss.

关键词： Co-attention Multimodal Interaction VQA

来源：评论

学校读者我要写书评

暂无评论

indian Regional Sign Language Recognition 23

Indian Regional Sign Language Recognition

引用

proceedings of the Fourteenth indian conference on computer vision, graphics and image processing

作者： Sujatha C Madhura K Patil Umadevi. F.M Neha Jain Akshata Badni Ranjeeta Angadi Shankar Gangisetty Computer Science and Engineering KLE Technological University INDIA Computer Science and Engineering KLE Technological University Hubballi INDIA Computer Science and Engineering IIIT Hyderabad INDIA

ISBN: (纸本)9798400716256

Sign Language is a medium of communication in the Deaf and Hard of Hearing community (DHH community). According to WHO, there are approximately 63 million people in India, including 3.3 million from the state of Karnataka, with 0.3 million children belonging to DHH community. The sign language gestures vary globally, making communication difficult between the DHH community and others. Hence, Sign Language Recognition has gained significant attention in recent years. There is a limited publicly available dataset on indian Sign Language. In addition, the dataset for regional languages especially the Kannada dataset, is unavailable for the sign recognition tasks. Hence initially, in our work, we built a dataset of indian Regional Kannada Sign Language (IRKSL) for different levels of data at character, word, and sentence levels. We propose a method for recognizing the indian sign language and translating it into Kannada regional language. We demonstrate the proposed method on the IRKSL dataset and obtained an accuracy of 81.69% and 73.87% with and without feature augmentation, respectively, on the IRKSL dataset. We also evaluated our work on the INCLUDE 50 dataset and obtained a better accuracy of 93.42% as compared to the state-of-art method.

关键词： Attention based BiLSTM CNN Feature Augmentation indian Sign Language Recognition

来源：评论

学校读者我要写书评

暂无评论

Identification of Medicinal Plants in Ardabil Using Deep learning: Identification of Medicinal Plants using Deep learning 27

Identification of Medicinal Plants in Ardabil Using Deep lea...

引用

27th International computer conference, computer Society of Iran, CSICC 2022

作者： Abdollahi, Jafar Islamic Azad University Ardabil Branch Department Of Computer Engineering Ardabil Iran

ISBN: (纸本)9781665480277

Ardabil is well-known for offering the ideal environment for a good, cheap medicinal herb. Various plant parts are used as essential components in producing natural medicines. According to IUCN (International Union for Conservation of Nature) records, many medicinal plants are on the verge of extinction, so employing image processing and computer vision algorithms to distinguish proof of medicinal plants is critical. As a result, the digitalization of beneficial therapeutic plants is critical for biodiversity preservation. The use of Convolutional Neural Network (CNN)-based techniques to distinguish indian leaf species is investigated in this research. Several Deep Learning frameworks have recently been used to discern, identify, and characterize various plants. This study is mostly focused on identifying medicinal plants that can be found in rural areas. The Transfer Learning technique selected a well-known pre-trained CNN architecture called mobile net v2. The medical plant dataset was built using 30 different classes of medicinal plants, totaling 3000 photos, and these models were assessed with their pre-trained weights. On a held-out test set, the trained model had an accuracy of 98.05 percent, demonstrating the practicality of this approach. © 2022 IEEE.

关键词： Deep learning computer vision image processing Transfer learning Knowledge discovery Software Data systems

来源：评论

学校读者我要写书评

暂无评论

End-to-End GPU-Accelerated Low-Poly Remeshing using Curvature Map and Voronoi Tessellationg± 13

End-to-End GPU-Accelerated Low-Poly Remeshing using Curvatur...

引用

13th indian conference on computer vision, graphics, and image processing, ICVGIP 2022

作者： Chatterjee, Preetam Chayan Soni, Ashutosh Bhowmick, Partha Bhowmick, Iit Kharagpur Indian Institute of Technology West Bengal Kharagpur India ANSYS Software Pvt. Ltd Hinjewadi Maharashtra Pune India

ISBN: (纸本)9781450398237

We propose a novel algorithm for low-poly remeshing of 3D surfaces that runs fully in GPU. Since the input mesh is generally not well-organized, performing mesh simplification directly on the input mesh is liable to produce a low-poly mesh with a compromised face quality. Hence, instead of doing that, we voxelize the input mesh first. On the voxelized surface, we perform curvature estimation and discretization, Centroidal Voronoi Tessellation (CVT), and Delaunay triangulation to generate the output mesh. To control the vertex count, we exploit the curvature map, such that sufficiently large triangles constitute the low-curvature regions, whereas appropriately small triangles build up the high-curvature portions. We produce low-poly meshes at various levels of detail and measure their quality using Hausdorff error. We test our algorithm on different 3D scenes, and they are particularly chosen because they contain multiple objects with varied topologies. Upon testing on these datasets, our algorithm successfully achieves very low-poly approximation with a reasonable mesh quality, and at a reasonably faster pace, thus establishing its efficacy for mesh simplification. © 2022 ACM.

关键词： graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

ContextCLIP: Contextual Alignment of image-Text pairs on CLIP visual representations 13

ContextCLIP: Contextual Alignment of Image-Text pairs on CLI...

引用

13th indian conference on computer vision, graphics, and image processing, ICVGIP 2022

作者： Grover, Chanda Mastan, Indra Deep Gupta, Debayan Ashoka University Sonipat Haryana India Rajasthan Jaipur India

ISBN: (纸本)9781450398237

State-of-The-Art empirical work has shown that visual representations learned by deep neural networks are robust in nature and capable of performing classification tasks on diverse datasets. For example, CLIP demonstrated zero-shot transfer performance on multiple datasets for classification tasks in a joint embedding space of image and text pairs. However, it showed negative transfer performance on standard datasets, e.g., BirdsNAP, RESISC45, and MNIST. In this paper, we propose ContextCLIP, a contextual and contrastive learning framework for the contextual alignment of image-Text pairs by learning robust visual representations on Conceptual Captions dataset. Our framework was observed to improve the image-Text alignment by aligning text and image representations contextually in the joint embedding space. ContextCLIP showed good qualitative performance for text-To-image retrieval tasks and enhanced classification accuracy. We evaluated our model quantitatively with zero-shot transfer and fine-Tuning experiments on CIFAR-10, CIFAR-100, Birdsnap, RESISC45, and MNIST datasets for classification task. © 2022 ACM.

关键词： Embeddings

来源：评论

学校读者我要写书评

暂无评论

Variational Augmentation for Enhancing Historical Document image Binarization✱ 22

Variational Augmentation for Enhancing Historical Document I...

引用

proceedings of the Thirteenth indian conference on computer vision, graphics and image processing

作者： Avirup Dey Prof. Mita Nasipuri Nibaran Das Jadavpur University India

ISBN: (纸本)9781450398220

Historical Document image Binarization is a well-known seg- mentation problem in image processing. Despite ubiquity, traditional thresholding algorithms achieved limited success on severely degraded document images. With the advent of deep learning, several segmentation models were proposed that made significant progress in the field but were limited by the unavailability of large training datasets. To mitigate this problem, we have proposed a novel two-stage framework- the first of which comprises a generator that generates degraded samples using variational inference and the second being a CNN-based binarization network that trains on the generated data. We evaluated our framework on a range of DIBCO datasets, where it achieved competitive results against previous state-of-the-art methods.

关键词： Binarization

来源：评论

学校读者我要写书评

暂无评论

Efficient-PSP-Net: A Lightweight Model for the Segmentation of Liver and Tumor in CT Scans 13

Efficient-PSP-Net: A Lightweight Model for the Segmentation ...

引用

13th indian conference on computer vision, graphics, and image processing, ICVGIP 2022

作者： Sindhura, Chitimireddy Vengama, Harsha Vardhan Reddy Gorthi, Subrahmanyam Indian Institute of Technology Tirupati Andhra Pradesh India

ISBN: (纸本)9781450398237

The segmentation of the liver and tumors present in the liver is an essential task for the diagnosis of liver cancer and surgical treatment planning. While most of the works were developed with the goal of improving segmentation accuracy, there is a high demand for the development of lightweight architectures to deploy in clinical settings with limited computational resources. Developing such resource-efficient models while maintaining the performance of the state-of-The-Art architectures is challenging due to the high class imbalance and small observable changes between tumorous and non-Tumorous regions, which would generally require extensive and complex deep learning architectures. The contributions of this paper are twofold;First, this paper proposes to encode information extracted at different scales using a "Pyramid Pooling Module"(PPM) architecture. While different variants of pyramidal architectures are widely used in the context of scene parsing in computer vision, this paper explores the use of the PPM architecture for the segmentation of liver and tumor. Second, the paper proposes using the popular lightweight "Efficient-Net"as the backbone network of PSP-Net for feature extraction, thereby reducing the overall model size while achieving relatively similar performance to state-of-The-Art architectures. The performance of the proposed model is evaluated on the MICCAI 2017 LiTS dataset. The model achieved a DSC of 95.06% and 79.08% for liver and tumor, respectively, with only 1.76M parameters. © 2022 ACM.

关键词： Tumors

来源：评论

学校读者我要写书评

暂无评论

Enhanced Kernelized Conditional Spatial Fuzzy C Means Algorithm for Noisy Brain MRI Tissue Segmentation 13

Enhanced Kernelized Conditional Spatial Fuzzy C Means Algori...

引用

13th indian conference on computer vision, graphics, and image processing, ICVGIP 2022

作者： Halder, Amiya Choudhuri, Rudrajit Sarkar, Apurba St Thomas College of Engineering and Technology West Bengal Kolkata India IIEST Shibpur Howrah West Bengal India

ISBN: (纸本)9781450398237

This paper proposes an Enhanced Kernelized Conditional Spatial Fuzzy C Means (EKCSFCM) clustering algorithm targeted at segmentation of Brain Magnetic Resonance (MR) image data. The approach amalgamates kernel mapping with a spatially relevant fuzzy clustering paradigm for image segmentation. A technique for rank-based consideration of spatial information is introduced and is used along with spatial tuning parameters for controlling neighborhood effects during data point clustering. The kernelized approach and the spatial parameters handle intensity inhomogeneities and strengthen the performance of the algorithm by improving resiliency against noise and other aberrations. For analyzing the robustness of the approach, four volumes of brain MR data, each comprising 71 images are used. These images are also corrupted with noise and inhomogeneities for evaluating the efficiency of different approaches in noisy image segmentation. The experimentation results along with the qualitative and quantitative inferences verify the robustness of the algorithm across a wide variety of test cases. © 2022 ACM.

关键词： image segmentation

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：