Ear Detection and Recognition Techniques: A Comparative Review

Srivastava, Pallavi; Agrawal, Diwakar; Bansal, Atul

doi:10.1007/978-981-15-0694-9_50

Cited by 9 publications

(2 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Kulak tespiti ise herhangi bir eşleşme olup olmadığının aranması ve bir eşleşme bulunduğunda arama işleminin durdurulmasıdır [19]. Literatürde, kulak tespiti ve tanıma üzerine yapılan birçok klasik makine öğrenmesi ve derin öğrenme çalışmaları vardır [20][21][22]. Alshazly vd.…”

Section: Kulak Tanıma Ve Kulak Tespiti (Ear Recognition and Ear Detec...unclassified

See 1 more Smart Citation

İnsan kulağı görüntüleri kullanarak cinsiyet tanıma için derin öğrenme tabanlı melez bir yaklaşım

Karasulu

Yücalar

BORANDAG

2022

Gazi Üniversitesi Mühendislik Mimarlık Fakültesi Dergisi

View full text Add to dashboard Cite

 Gender recognition was made using images of human ears  Gender recognition based on the hybrid deep learning architectural models is presented to the literature  The success of the proposed approach has been shown using various human ear image datasets A hybrid deep learning approach is presented in our study to recognize more accurately the human gender with the use of ear images obtained unconditionally (i.e., in-the-wild) with various scale, orientation and size.On the contrary of classical approaches, it is seen by our hybrid approach that the requirements expected from any biometric system can be met by bridging the semantic gaps between lower-level and higher-level features thanks to the learned ear image attributes obtained with the use of deep learning. The main contribution of this study to the literature is to present the experimental results of both the convolutional neural network (CNN) model (as standalone) and a model based on the novel hybridization approach. Hybrid model was constructed with CNN component and the component with recurrent neural network (RNN)-type layers. Consequently, it was tested for classification (i.e., female or male recognition). Figure A. Schematic for gender recognition based on hybrid deep learning with human ear images.Purpose: This study aims to proof the necessity and capabilities of proposed our novel hybrid deep learning approach for gender recognition using human ear images. The study presents the performance benchmarking results of our gender recognition approach with the use of two different datasets with various scale, orientation and image sizes that these images were tagged with gender mark came from their datasets. Theory and Methods:To obtain hybridization, proposed approach uses convolutional neural network (CNN) model and recurrent neural network (RNN) models such as gated recurrent unit (GRU) and long-short term memory (LSTM). Our hybridization approach was tested by using the combination of these abovementioned deep learning architectural models to proof the human gender recognition performance with ear images. Results:In this study, the human gender classification and recognition performance achieved by the hybridization of deep learning architectural models varies with parameter configuration of these models in given experiments. By considering performance benchmark on ear images from two different datasets, the best test accuracy rates with various combination of these models were obtained 85.16% for EarVN (i.e., Vietnam) dataset and 87.61% for WPUT (i.e., Poland) dataset as well. Conclusion:Gender recognition was made on higher-level abstract features obtained throughout the representational learning of the hybrid deep neural network model working on the various images. Alongside the CNN component, thanks to components with RNN-type (e.g., LSTM and GRU) layers, a better understanding of the relational dependencies between pixel regions in the ear images for hybrid neural network model has been provided. The two-component hybrid deep neural network...

show abstract

Section: Kulak Tanıma Ve Kulak Tespiti (Ear Recognition and Ear Detec...unclassified

“…Kulak büyümesi orantılı olsa da tıbbi literatüre göre yerçekimi kulağın dikey yönde gerilmesine neden olabilmektedir [22]. Kulak yapısındaki oranlar yaş ilerledikçe asla değişmemektedir.…”

Section: Kulak Biyometrisi (Ear Biometry)unclassified