A Survey on Image Classification and Activity Recognition using Deep Convolutional Neural Network Architecture

Sornam, M.; Muthusubash, Kavitha; Vanitha, V.

doi:10.1109/icoac.2017.8441512

Cited by 40 publications

(11 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The F(2×2, 5×5) Winograd Transform is given in a previous work eq. (4). Table 2 shows the comparison between the number of operations used for the F(2 × 2, 5 × 5) Winograd Transform, spatial convolution and their corresponding energy costs.…”

Section: B Winograd Transformmentioning

confidence: 99%

See 1 more Smart Citation

TiQSA: Workload Minimization in Convolutional Neural Networks Using Tile Quantization and Symmetry Approximation

et al. 2021

View full text Add to dashboard Cite

Convolutional Neural Networks (CNNs) in the Internet-of-Things (IoT)-based applications face stringent constraints, like limited memory capacity and energy resources due to many computations in convolution layers. In order to reduce the computational workload in these layers, this paper proposes a hybrid convolution method in conjunction with a Particle of Swarm Convolution Layer Optimization (PSCLO) algorithm. The hybrid convolution is an approximation that exploits the inherent symmetry of filter termed as symmetry approximation and Winograd algorithm structure termed as tile quantization approximation. PSCLO optimizes the balance between workload reduction and accuracy degradation for each convolution layer by selecting fine-tuned thresholds to control each approximation's intensity. The proposed methods have been evaluated on ImageNet, MNIST, Fashion-MNIST, SVHN, and CIFAR-10 datasets. The proposed techniques achieved ∼5.28x multiplicative workload reduction without significant accuracy degradation (<0.1%) for ImageNet on ResNet-18, which is ∼1.08x less multiplicative workload as compared to state-of-the-art Winograd CNN pruning. For LeNet, ∼3.87x and ∼3.93x was the multiplicative workload reduction for MNIST and Fashion-MNIST datasets. The additive workload reduction was ∼2.5x and ∼2.56x for the respective datasets. There is no significant accuracy loss for MNIST and Fashion-MNIST dataset.

show abstract

Section: B Winograd Transformmentioning

confidence: 99%

“…During the last decade, deep learning algorithms for image classification tasks [1]- [3] progressively evolved towards deeper and more complex networks [4]- [7]. These approaches [8] are considered as effective in the field of image processing [9], signal processing [10] medical imaging [11], and speech recognition [12].…”

Section: Introductionmentioning

confidence: 99%

TiQSA: Workload Minimization in Convolutional Neural Networks Using Tile Quantization and Symmetry Approximation

et al. 2021

View full text Add to dashboard Cite

show abstract

“…The CNNs are widely used models for time series and image data classification [43][44][45][46]. There are different variants of CNN proposed in the literature such as AlexNet [47] ResNet [48], VGGnet [49], SqueezeNet [50] and GoogLeNet [51].…”

Section: = Selection Of the Input Mapsmentioning

confidence: 99%

Deep Learning Algorithms based Fingerprint Authentication: Systematic Literature Review

Chiroma¹

2021

AIS

View full text Add to dashboard Cite

Deep Learning algorithms (DL) have been applied in different domains such as computer vision, image detection, robotics and speech processing, in most cases, DL demonstrated better performance than the conventional machine learning algorithms (shallow algorithms). The artificial intelligence research community has leveraged the robustness of the DL because of their ability to process large data size and handle variations in biometric data such as aging or expression problem. Particularly, DL research in automatic fingerprint recognition system (AFRS) is gaining momentum starting from the last decade in the area of fingerprint pre-processing, fingerprints quality enhancement, fingerprint feature extraction, security of fingerprint and performance improvement of AFRS. However, there are limited studies that address the application of DL to model fingerprint biometric for different tasks in the fingerprint recognition process. To bridge this gap, this paper presents a systematic literature review and an insightful meta-data analysis of a decade applications of DL in AFRS. Discussion on proposed model’s tasks, state of the art study, dataset, and training architecture are presented. The Convolutional Neural Networks models were the most saturated models in developing fingerprint biometrics authentication. The study revealed different roles of the DL in training architecture of the models: feature extractor, classifier and end-to-end learning. The review highlights open research challenges and present new perspective for solving the challenges in the future. The author believed that this paper will guide researchers in propose novel fingerprint authentication scheme.

show abstract

“…Also, a combination of the Proposal-based technique and CNN was successful for image segmentation [55]. However, recently several CNN-based architectures have been developed [56], such that in some cases using only CNN is good enough for feature extraction and image analysis, such as image segmentation [37], [57] and classification [58].…”

Section: E Image Segmentationmentioning

confidence: 99%

An Evolutionary DenseRes Deep Convolutional Neural Network for Medical Image Segmentation

2020

View full text Add to dashboard Cite

The performance of a Convolutional Neural Network (CNN) highly depends on its architecture and corresponding parameters. Manually designing a CNN is a time-consuming process in regards to the various layers that it can have, and the variety of parameters that must be set up. Increasing the complexity of the network structure by employing various types of connections makes designing a network even more challenging. Evolutionary computation as an optimisation technique can be applied to arrange the CNN layers and/or initiate its parameters automatically or semi-automatically. Dense network and Residual network are two popular network structures that were introduced to facilitate the training of deep networks. In this paper, leveraging the potentials of Dense and Residual blocks, and using the capability of evolutionary computation, we propose an automatic evolutionary model to detect an optimum and accurate network structure and its parameters for medical image segmentation. The proposed evolutionary DenseRes model is employed for segmentation of six publicly available MRI and CT medical datasets. The proposed model obtained high accuracy while employing networks with minimal parameters for the segmentation of medical images and outperformed manual and automatic designed networks, including U

show abstract

A Survey on Image Classification and Activity Recognition using Deep Convolutional Neural Network Architecture

Cited by 40 publications

References 14 publications

TiQSA: Workload Minimization in Convolutional Neural Networks Using Tile Quantization and Symmetry Approximation

TiQSA: Workload Minimization in Convolutional Neural Networks Using Tile Quantization and Symmetry Approximation

Deep Learning Algorithms based Fingerprint Authentication: Systematic Literature Review

An Evolutionary DenseRes Deep Convolutional Neural Network for Medical Image Segmentation

Contact Info

Product

Resources

About