检索结果-内蒙古大学图书馆

Face to Cartoon Incremental Super-Resolution using Knowledge Distillation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Devkatte, Trinetra Dubey, Shiv Ram Singh, Satish Kumar Hadid, Abdenour Computer Vision and Biometrics Lab Indian Institute of Information Technology Allahabad India Sorbonne Centre for Artificial Intelligence Sorbonne University Abu Dhabi United Arab Emirates

Facial super-resolution/hallucination is an important area of research that seeks to enhance low-resolution facial images for a variety of applications. While Generative Adversarial Networks (GANs) have shown promise in this area, their ability to adapt to new, unseen data remains a challenge. This paper addresses this problem by proposing an incremental super-resolution using GANs with knowledge distillation (ISR-KD) for face to cartoon. Previous research in this area has not investigated incremental learning, which is critical for real-world applications where new data is continually being generated. The proposed ISR-KD aims to develop a novel unified framework for facial super-resolution that can handle different settings, including different types of faces such as cartoon face and various levels of detail. To achieve this, a GAN-based super-resolution network was pre-trained on the CelebA dataset and then incrementally trained on the iCartoonFace dataset, using knowledge distillation to retain performance on the CelebA test set while improving the performance on iCartoonFace test set. Our experiments demonstrate the effectiveness of knowledge distillation in incrementally adding capability to the model for cartoon face super-resolution while retaining the learned knowledge for facial hallucination tasks in GANs. Copyright © 2024, The Authors. All rights reserved.

关键词： Generative adversarial networks

Point-GR: Graph Residual Point Cloud Network for 3D Object Classification and Segmentation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Meraz, Md. Ansari, Md Afzal Javed, Mohammed Chakraborty, Pavan Center of Intelligent Robotics Lab Department of Information Technology Computer Vision & Biometrics Lab Department of Information Technology Indian Institute of Information Technology Allahabad U.P Prayagraj India

In recent years, the challenge of 3D shape analysis within point cloud data has gathered significant attention in computer vision. Addressing the complexities of effective 3D information representation and meaningful feature extraction for classification tasks remains crucial. This paper presents Point-GR, a novel deep learning architecture designed explicitly to transform unordered raw point clouds into higher dimensions while preserving local geometric features. It introduces residual-based learning within the network to mitigate the point permutation issues in point cloud data. The proposed Point-GR network significantly reduced the number of network parameters in Classification and PartSegmentation compared to baseline graph-based networks. Notably, the Point-GR model achieves a state-of-the-art scene segmentation mean IoU of 73.47% on the S3DIS benchmark dataset, showcasing its effectiveness. Furthermore, the model shows competitive results in Classification and Part-Segmentation tasks. © 2024, CC BY.

关键词： Graph neural networks

Non-Uniform Illumination Attack for Fooling Convolutional Neural Networks

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Jain, Akshay Dubey, Shiv Ram Singh, Satish Kumar Santosh, K.C. Chaudhuri, Bidyut Baran The Computer Vision and Biometrics Lab Department of Information Technology Indian Institute of Information Technology Allahabad Uttar Pradesh Prayagraj211015 India The AI Research Lab Department of Computer Science University of South Dakota VermillionSD57069 United States The Computer Vision and Pattern Recognition Unit Indian Statistical Institute Kolkata700108 India

Convolutional Neural Networks (CNNs) have made remarkable strides;however, they remain susceptible to vulnerabilities, particularly in the face of minor image perturbations that humans can easily recognize. This weakness, often termed as ‘attacks,’ underscores the limited robustness of CNNs and the need for research into fortifying their resistance against such manipulations. This study introduces a novel Non-Uniform Illumination (NUI) attack technique, where images are subtly altered using varying NUI masks. Extensive experiments are conducted on widely-accepted datasets including CIFAR10, TinyImageNet, and CalTech256, focusing on image classification with 12 different NUI attack models. The resilience of VGG, ResNet, MobilenetV3-small and InceptionV3 models against NUI attacks are evaluated. Our results show a substantial decline in the CNN models’ classification accuracy when subjected to NUI attacks, indicating their vulnerability under non-uniform illumination. To mitigate this, a defense strategy is proposed, including NUI-attacked images, generated through the new NUI transformation, into the training set. The results demonstrate a significant enhancement in CNN model performance when confronted with perturbed images affected by NUI attacks. This strategy seeks to bolster CNN models’ resilience against NUI attacks. 1 Copyright © 2024, The Authors. All rights reserved.

关键词： Convolutional neural networks

Alt4Blind: A User Interface to Simplify Charts Alt-Text Creation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Moured, Omar Farooqui, Shahid Ali Müller, Karin Fadaeijouybari, Sharifeh Schwarz, Thorsten Javed, Mohammed Stiefelhagen, Rainer CV:HCI@KIT Karlsruhe Institute of Technology Germany ACCESS@KIT Karlsruhe Institute of Technology Germany Computer Vision & Biometrics Lab Indian Institute of Information Technology Allahabad India University of Trier Germany

Alternative Texts (Alt-Text) for chart images are essential for making graphics accessible to people with blindness and visual impairments. Traditionally, Alt-Text is manually written by authors but often encounters issues such as oversimplification or complication. Recent trends have seen the use of AI for Alt-Text generation. However, existing models are susceptible to producing inaccurate or misleading information. We address this challenge by retrieving high-quality alt-texts from similar chart images, serving as a reference for the user when creating alt-texts. Our three contributions are as follows: (1) we introduce a new benchmark comprising 5,000 real images with semantically labeled high-quality Alt-Texts, collected from Human computer Interaction venues. (2) We developed a deep learning-based model to rank and retrieve similar chart images that share the same visual and textual semantics. (3) We designed a user interface (UI) to facilitate the alt-text creation process. Our preliminary interviews and investigations highlight the usability of our UI. For the dataset and further details, please refer to our project page: https://***/alt4blind/. © 2024, CC BY.

关键词： Image retrieval

Target Aware Network Architecture Search and Compression for Efficient Knowledge Transfer

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Basha, S.H. Shabbeer Tula, Debapriya Vinakota, Sravan Kumar Dubey, Shiv Ram RV University Karnataka Bengaluru India Google Research Karnataka Bangalore India New Jersey Institute of Technology NJ United States Computer Vision and Biometrics Lab Indian Institute of Information Technology Allahabad Uttar Pradesh Prayagraj India

Transfer Learning enables Convolutional Neural Networks (CNN) to acquire knowledge from a source domain and transfer it to a target domain, where collecting large-scale annotated examples is time-consuming and expensive. Conventionally, while transferring the knowledge learned from one task to another task, the deeper layers of a pre-trained CNN are finetuned over the target dataset. However, these layers are originally designed for the source task which may be over-parameterized for the target task. Thus, finetuning these layers over the target dataset may affect the generalization ability of the CNN due to high network complexity. To tackle this problem, we propose a two-stage framework called TASCNet which enables efficient knowledge transfer. In the first stage, the configuration of the deeper layers is learned automatically and finetuned over the target dataset. Later, in the second stage, the redundant filters are pruned from the fine-tuned CNN to decrease the network’s complexity for the target task while preserving the performance. This two-stage mechanism finds a compact version of the pre-trained CNN with optimal structure (number of filters in a convolutional layer, number of neurons in a dense layer, and so on) from the hypothesis space. The efficacy of the proposed method is evaluated using VGG-16, ResNet-50, and DenseNet-121 on CalTech-101, CalTech-256, and Stanford Dogs datasets. Similar to computer vision tasks, we have also conducted experiments on Movie Review Sentiment Analysis task. The proposed TASCNet reduces the computational complexity of pre-trained CNNs over the target task by reducing both trainable parameters and FLOPs which enables resource-efficient knowledge transfer. The source code is available at: https://***/Debapriya-Tula/TASCNet. Copyright © 2022, The Authors. All rights reserved.

关键词： Network architecture

LiSHT: Non-Parametric Linearly Scaled Hyperbolic Tangent Activation Function for Neural Networks

学校读者我要写书评

暂无评论

arXiv 2018年

作者： Roy, Swalpa Kumar Manna, Suvojit Dubey, Shiv Ram Chaudhuri, Bidyut Baran Department of Computer Science & Engineering Jalpaiguri Government of Engineering College West Bengal India CureSkin Karnataka Bengaluru India Computer Vision and Biometrics Lab Indian Institute of Information Technology Uttar Pradesh Allahabad India Techno India University Kolkata India Indian Statistical Institute Kolkata India

The activation function in neural network introduces the non-linearity required to deal with the complex tasks. Several activation/non-linearity functions are developed for deep learning models. However, most of the existing activation functions suffer due to the dying gradient problem and non-utilization of the large negative input values. In this paper, we propose a Linearly Scaled Hyperbolic Tangent (LiSHT) for Neural Networks (NNs) by scaling the Tanh linearly. The proposed LiSHT is non-parametric and tackles the dying gradient problem. We perform the experiments on benchmark datasets of different type, such as vector data, image data and natural language data. We observe the superior performance using Multi-layer Perceptron (MLP), Residual Network (ResNet) and Long-short term memory (LSTM) for data classification, image classification and tweets classification tasks, respectively. The accuracy on CIFAR100 dataset using ResNet model with LiSHT is improved by 9.48, 3.40, 3.16, 4.26, and 1.17% as compared to Tanh, ReLU, PReLU, LReLU, and Swish, respectively. We also show the qualitative results using loss landscape, weight distribution and activations maps in support of the proposed activation function. Copyright © 2018, The Authors. All rights reserved.

关键词： Image classification

Frequency disentangled residual network

学校读者我要写书评

暂无评论

Multimedia Systems 2024年第1期30卷 9-9页

作者： Singh, Satya Rajendra Yedla, Roshan Reddy Dubey, Shiv Ram Sanodiya, Rakesh Kumar Chu, Wei-Ta Computer Science and Engineering Group Indian Institute of Information Technology Andhra Pradesh Sri India Stony Brook University Stony BrookNY United States Computer Vision and Biometrics Lab Indian Institute of Information Technology Uttar Pradesh Allahabad India Department of Computer Science and Information Engineering National Cheng Kung University Tainan Taiwan

Residual networks (ResNets) have been utilized for various computer vision and image processing applications. The residual connection improves the training of the network with better gradient flow. A residual block consists of a few convolutional layers having trainable parameters, which leads to overfitting. Moreover, the present residual networks are not able to utilize the high- and low-frequency information suitably, which also challenges the generalization capability of the network. In this paper, a frequency-disentangled residual network (FDResNet) is proposed to tackle these issues. Specifically, FDResNet includes separate connections in the residual block for low- and high-frequency components, respectively. Basically, the proposed model disentangles the low- and high-frequency components to increase the generalization ability. Moreover, the computation of low- and high-frequency components using fixed filters further avoids the overfitting. The proposed model is tested on benchmark CIFAR-10/100, Caltech, and TinyImageNet datasets for image classification. The performance of the proposed model is also tested in the image retrieval framework. It is noticed that the proposed model outperforms its counterpart residual model. The effect of kernel size and standard deviation is also evaluated. The impact of the frequency disentangling is also analyzed using a saliency map. © 2024, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.

关键词： Image classification

AngularGrad: A New Optimization Technique for Angular Convergence of Convolutional Neural Networks

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Roy, Swalpa Kumar Paoletti, Mercedes E. Haut, Juan M. Dubey, Shiv Ram Kar, Purbayan Plaza, Antonio Chaudhuri, Bidyut B. The Computer Science and Engineering Alipurduar Government Engineering and Management College 736206 India The Hyperspectral Computing Laboratory Department of Technology of Computers and Communications University of Extremadura Cáceres10003 Spain The Computer Vision and Biometrics Lab Indian Institute of Information Technology Prayagraj Uttar Pradesh Allahabad211015 India The Media Analysis Group Sony Research India Private Limited Karnataka Bangalore560103 India The Computer Vision and Pattern Recognition Unit Indian Statistical Institute Kolkata700108 India

Convolutional neural networks (CNNs) are trained using stochastic gradient descent (SGD)-based optimizers. Recently, the adaptive moment estimation (Adam) optimizer has become very popular due to its adaptive momentum, which tackles the dying gradient problem of SGD. Nevertheless, existing optimizers are still unable to exploit the optimization curvature information efficiently. This paper proposes a new AngularGrad optimizer that considers the behavior of the direction/angle of consecutive gradients. This is the first attempt in the literature to exploit the gradient angular information apart from its magnitude. The proposed AngularGrad generates a score to control the step size based on the gradient angular information of previous iterations. Thus, the optimization steps become smoother as a more accurate step size of immediate past gradients is captured through the angular information. Two variants of AngularGrad are developed based on the use of Tangent or Cosine functions for computing the gradient angular information. Theoretically, AngularGrad exhibits the same regret bound as Adam for convergence purposes. Nevertheless, extensive experiments conducted on benchmark data sets against state-of-the-art methods reveal a superior performance of AngularGrad. Source code: https://***/mhaut/AngularGrad. Copyright © 2021, The Authors. All rights reserved.

关键词： Optimization

EMAHA-DB1: A New Upper Limb sEMG Dataset for Classification of Activities of Daily Living

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Karnam, Naveen Kumar Turlapaty, Anish Chand Dubey, Shiv Ram Gokaraju, Balakrishna The Biosignal Analysis Lab The Indian Institute of Information Technology A.P. Sri City India The Computer Vision and Biometrics Laboratory Indian Institute of Information Technology U.P. Allahabad Prayagraj211015 India Department of Computational Data Science An Engineering North Carolina A and T State University GreensboroNC United States

In this paper, we present electromyography analysis of human activity - database 1 (EMAHA-DB1), a novel dataset of multi-channel surface electromyography (sEMG) signals to evaluate the activities of daily living (ADL). The dataset is acquired from 25 able-bodied subjects while performing 22 activities categorised according to functional arm activity behavioral system (FAABOS) (3 - full hand gestures, 6 - open/close office draw, 8 - grasping and holding of small office objects, 2 - flexion and extension of finger movements, 2 - writing and 1 - rest). The sEMG data is measured by a set of five Noraxon Ultium wireless sEMG sensors with Ag/Agcl electrodes placed on a human hand. The dataset is analyzed for hand activity recognition classification performance. The classification is performed using four state-of-the-art machine learning classifiers, including Random Forest (RF), Fine K-Nearest Neighbour (KNN), Ensemble KNN (sKNN) and Support Vector Machine (SVM) with seven combinations of time domain and frequency domain feature sets. The state-of-the-art classification accuracy on five FAABOS categories is 83.21% by using the SVM classifier with the third order polynomial kernel using energy feature and auto regressive feature set ensemble. The classification accuracy on 22 class hand activities is 75.39% by the same SVM classifier with the log moments in frequency domain (LMF) feature, modified LMF, time domain statistical (TDS) feature, spectral band powers (SBP), channel cross correlation and local binary patterns (LBP) set ensemble. The analysis depicts the technical challenges addressed by the dataset. The developed dataset can be used as a benchmark for various classification methods as well as for sEMG signal analysis corresponding to ADL and for the development of prosthetics and other wearable robotics. Copyright © 2023, The Authors. All rights reserved.

关键词： Local binary pattern