Wavelet features embedded convolutional neural network for multiscale ear recognition

Mewada, Hiren; Patel, Amit; Chaudhari, Jitendra; Mahant, Keyur; Vala, Alpesh

doi:10.1117/1.jei.29.4.043029

Cited by 9 publications

(4 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Additionally, the authors proposed a deep unsupervised active learning-based ear recognition system [19], which was tested in both controlled and uncontrolled conditions, further demonstrating the adaptability of deep learning techniques to diverse environments. Mewada et al [20] proposed a spectral-spatial feature based on CNN for describing ear images and an embedding algorithm for fusing multilevel spectral information from the CNN network. The performance of the proposed system was evaluated on ear images captured under uncontrolled conditions.…”

Section: Related Workmentioning

confidence: 99%

Deep Learning Based Gender Identification Using Ear Images

Kılıç,

Doğan

2023

View full text Add to dashboard Cite

The classification of an individual as male or female is a significant issue with several practical implications. In recent years, automatic gender identification has garnered considerable interest because of its potential applications in e-commerce and the accumulation of demographic data. Recent observations indicate that models based on deep learning have attained remarkable success in a variety of problem domains. In this study, our aim is to establish an end-to-end model that capitalizes on the strengths of competing convolutional neural network (CNN) and vision transformer (ViT) models. To accomplish this, we propose a novel approach that combines the MobileNetV2 model, which is recognized for having fewer parameters than other CNN models, with the ViT model. Through rigorous evaluations, we have compared our proposed model with other recent studies using the accuracy metric. Our model attained state-of-the-art performance with a remarkable score of 96.66% on the EarVN1.0 dataset, yielding impressive results. In addition, we provide t-SNE results that demonstrate our model's superior learning representation. Notably, the results show a more effective disentanglement of classes.

show abstract

Section: Related Workmentioning

confidence: 99%

Deep Learning Based Gender Identification Using Ear Images

Kılıç,

Doğan

2023

View full text Add to dashboard Cite

show abstract

“…Mel frequency cepstral coefficients (MFCCs), linear prediction coefficients (LPCs), linear prediction cepstral coefficients (LPCCs), line spectral frequencies (LSFs), discrete wavelet transform (DWT) [3,4], and perceptual linear prediction (PLP) are speech feature extractions commonly used in speaker recognition as well as speaker spoofing identification [5]. A wavelet transform was used to obtain spectral features, and these features were integrated with CNN's spatial features in Reference [6] for ECG classification.…”

Section: Introductionmentioning

confidence: 99%

Gaussian-Filtered High-Frequency-Feature Trained Optimized BiLSTM Network for Spoofed-Speech Classification

Mewada,

Al-Asad,

Almalki

et al. 2023

Sensors

View full text Add to dashboard Cite

Voice-controlled devices are in demand due to their hands-free controls. However, using voice-controlled devices in sensitive scenarios like smartphone applications and financial transactions requires protection against fraudulent attacks referred to as “speech spoofing”. The algorithms used in spoof attacks are practically unknown; hence, further analysis and development of spoof-detection models for improving spoof classification are required. A study of the spoofed-speech spectrum suggests that high-frequency features are able to discriminate genuine speech from spoofed speech well. Typically, linear or triangular filter banks are used to obtain high-frequency features. However, a Gaussian filter can extract more global information than a triangular filter. In addition, MFCC features are preferable among other speech features because of their lower covariance. Therefore, in this study, the use of a Gaussian filter is proposed for the extraction of inverted MFCC (iMFCC) features, providing high-frequency features. Complementary features are integrated with iMFCC to strengthen the features that aid in the discrimination of spoof speech. Deep learning has been proven to be efficient in classification applications, but the selection of its hyper-parameters and architecture is crucial and directly affects performance. Therefore, a Bayesian algorithm is used to optimize the BiLSTM network. Thus, in this study, we build a high-frequency-based optimized BiLSTM network to classify the spoofed-speech signal, and we present an extensive investigation using the ASVSpoof 2017 dataset. The optimized BiLSTM model is successfully trained with the least epoch and achieved a 99.58% validation accuracy. The proposed algorithm achieved a 6.58% EER on the evaluation dataset, with a relative improvement of 78% on a baseline spoof-identification system.

show abstract

“…However, this system is tedious, sensitive to the acquisition process, and computationally complex. New emerging ear trait recognition is explained in [5]. Mewada et al [5] used CNN for learning and discriminating between ear features.…”

Section: Introductionmentioning

confidence: 99%

“…New emerging ear trait recognition is explained in [5]. Mewada et al [5] used CNN for learning and discriminating between ear features. A moving human leg angle-based gait recognition system, explained in [6], uses average statistical features of eight frames.…”

Section: Introductionmentioning

confidence: 99%

Computationless Palm-Print Verification Using Wavelet Oriented Zero-Crossing Signature

Chaudhari

Mewada²,

Patel

et al. 2022

IIUMEJ

Self Cite

View full text Add to dashboard Cite

Palmprints can be characterized by their texture and the patterns of that texture dominate in a vertical direction. Therefore, the energy of the coefficients in the transform domain is more concentrated in the vertical sideband. Using this idea, this paper proposes the characterization of the texture features of the palmprint using zero-crossing signatures based on the dyadic discrete wavelet transform (DWT) to effectively identify an individual. A zero-crossing signature of 4 x 256 was generated from the lower four resolution levels of dyadic DWT in the enrolment process and stored in the database to identify the person in recognition mode. Euclidean distance was determined to find the best fit for query palmprints zero-crossing signature from the dataset. The proposed algorithm was tested on the PolyU dataset containing 6000 multi-spectral images. The proposed algorithm achieved 96.27% accuracy with a lower recognition time of 0.76 seconds. ABSTRAK: Pengesan Tapak Tangan boleh dikategorikan berdasarkan ciri-ciri tekstur dan corak pada tekstur yang didominasi pada garis tegak. Oleh itu, pekali tenaga di kawasan transformasi adalah lebih penuh pada jalur-sisi menegak. Berdasarkan idea ini, cadangan kajian ini adalah berdasarkan ciri-ciri tekstur pada tapak tangan dan tanda pengenalan sifar-silang melalui transformasi gelombang kecil diadik yang diskret (DWT) bagi mengecam individu. Pada mod pengecaman, tanda pengenalan sifar-silang 4 x 256 yang terhasil daripada tahap diadik resolusi empat terendah DWT digunakan dalam proses kemasukan dan simpanan di pangkalan data bagi mengenal pasti individu. Jarak Euklidan yang terhasil turut digunakan bagi memperoleh padanan tapak tangan paling sesuai melalui tanda pengenalan sifar-silang dari set data. Algoritma yang dicadangkan ini diuji pada set data PolyU yang mengandungi 6000 imej pelbagai-spektrum. Algoritma yang dicadangkan ini berjaya mencapai ketepatan sebanyak 96.27% dengan durasi pengecaman berkurang sebanyak 0.76 saat.

show abstract

Wavelet features embedded convolutional neural network for multiscale ear recognition

Cited by 9 publications

References 0 publications

Deep Learning Based Gender Identification Using Ear Images

Deep Learning Based Gender Identification Using Ear Images

Gaussian-Filtered High-Frequency-Feature Trained Optimized BiLSTM Network for Spoofed-Speech Classification

Computationless Palm-Print Verification Using Wavelet Oriented Zero-Crossing Signature

Contact Info

Product

Resources

About