Figure 2 - uploaded by Shan Sung Liew
Content may be subject to copyright.
Source publication
An approach using a convolutional neural network (CNN) is proposed for real-time gender classification based on facial images. The proposed CNN architecture exhibits a much reduced design complexity when compared with other CNN solutions applied in pattern recognition. The number of processing layers in the CNN is reduced to only four by fusing the...
Contexts in source publication
Context 1
... CNN is a type of feedforward network structure that is formed by multiple layers of convolutional filters alternated with subsampling filters followed by fully connected layers. Figure 2 shows the classical LeNet-5 CNN, first introduced by LeCun et al. in [4], which is the basis of the design of conventional CNNs. In [4], it was successfully applied in a handwritten digit recognition problem. ...
Context 2
... has six processing layers, not including the input layer, which is of image size 32 × 32 pixels. As illustrated in Figure 2, the processing layers consist of three convolutional layers C1, C3, and C5 interspersed in between with two subsampling layers, S2 and S4, and an output layer, F6. The convolutional and subsampling layers are organized into planes called feature maps. ...
Similar publications
Un réseau de neurone artificiel est un modèle de calcul qui se repose sur les méthodes d’apprentissage. Ce modèle est inspiré du fonctionnement des neurones biologiques qui constituent le cerveau.
En informatique bio-inspiré, les réseaux de neurones sont classés parmi les techniques d’intelligence artificielle qui tente de résoudre les problèmes c...
The purported "black box"' nature of neural networks is a barrier to adoption in applications where interpretability is essential. Here we present DeepLIFT (Deep Learning Important FeaTures), a method for decomposing the output prediction of a neural network on a specific input by backpropagating the contributions of all neurons in the network to e...
We demonstrate the utility of machine learning algorithms for the design of Oscillatory Neural Networks (ONNs). After constructing a circuit model of the oscillators in a machine-learning-enabled simulator and performing Backpropagation through time (BPTT) for determining the coupling resistances between the ring oscillators, we show the design of...
In the quest to enhance the efficiency and bio-plausibility of training deep neural networks, Feedback Alignment (FA), which replaces the backward pass weights with random matrices in the training process, has emerged as an alternative to traditional backpropagation. While the appeal of FA lies in its circumvention of computational challenges and i...
In this study, a novel hybrid algorithm consisting of the least mean square and backpropagation neural network is proposed to auto-adjust adaptive proportional integral derivative (PID) controller gains for improving the transient response of linear systems. The hybrid approach comprises the scheme of the two algorithms running in parallel and upda...
Citations
... The results of this study obtained an accuracy of 96.29%. Liew et al. [18] classify gender by using CNN algorithm. This study used publicly available datasets namely SUM and AT&T. ...
Nutmeg seeds can produce a lot of oil if they have optimal maturity, in other words, they have little moisture content. Based on observations I made at one of the refineries in Sukabumi, farmers do not pay attention to the maturity level of nutmeg seeds after drying which can cause a decrease in the quality of nutmeg seeds and the quality of the oil produced. This study aims to make it easier for nutmeg farmers to classify the maturity of nutmeg seeds. This study used the Convolutional Neural Network (CNN) method to help with classification problems and several image processing methods. This program will be run through an android application. The results of CNN model training accuracy are 97.92%. Thus, it can be concluded that the design and testing of a model to classify the maturity level of nutmeg seeds using artificial intelligence and the implementation of the model into an android application has been successfully carried out.
... A convolutional neural network and deep learning were used to identify gender categorization. It took four layers to get an accuracy of 8.759 % [19]. This method has applications in forensic medicine, where it integrates face measuring for images with deep learning at a pace of more than 3700 images per second with an accuracy of 89 % [20]. ...
Large communications of voice and images over the Internet, which leads to limited space for very large data during the aforementioned correspondence, to overcome this issue to maintain the quality of this technology. The proposal in the efficient algorithm in this work is a method to derive the two new filters from the second and third Chebyshev polynomials by forming the discrete wavelets with the mother wavelet to be used in image processing in order to overcome the problem mentioned above due to the correspondence, The filters that were derived are Filter Discrete Second Chebyshev Wavelets Transform (FDSCWT) and Filter Discrete Third Chebyshev Wavelets Transform (FDTCWT)to process the image by analysis, noise reduction, and image compression. Many of the techniques previously used in the field of image processing do not preserve image information during processing, but when using the new technology proposed in this work, it has been proven to preserve the image with its important information and data through the readings obtained shown in the tables below. These readings are average. Mean Square Error (MSE), Peak Signal Noise Ratio (PSNR), Bits Per Pixel (BPP), and Compression Ratio (CR) in preprocessing. After the initial processing stage, the deep learning stage begins in the field of artificial intelligence. A (CNN) is trained with the two new filters to be the first Discrete Second Chebyshev Wavelets (DSCWCNN) and the second Discrete Third Chebyshev Wavelets Convolutional Neural Network (DTCWCNN), with the code being generated in the MATLAB program with a network Alex Net to complete the classification process that was added in this work to implement the recognition technology. Faces detection with new filters in deep learning to be a unique experience to reach a high level of accuracy of 98.60 % with the network for the filter DSCWCNN and 98.92 % with the network for the filter DTCWCNN in a very short time, which will be mentioned later in the work
... There have been numerous earlier investigations regarding gender classification based on facial images. Liew et al. used the Convolutional Neural Network to classify a gender [2]. Asmara et al. succeeded in classifying gender using the Naïve Bayes method [3]. ...
... There have been numerous investigations related to gender classification based on face images. Previously Liew et al. obtained 99.38% accuracy using the Convolutional Neural Network for such a classification [2]. Here, the authors used the AT&T face database dataset containing 400 facial images, each represented in 32 32 pixels. ...
Due to various factors that cause visual alterations in the collected facial images, gender classification based on image processing continues to be a performance challenge for classifier models. The Vision Transformer model is used in this study to suggest a technique for identifying a person’s gender from their face images. This study investigates how well a facial image-based model can distinguish between male and female genders. It also investigates the rarely discussed performance on the variation and complexity of data caused by differences in racial and age groups. We trained on the AFAD dataset and then carried out same-dataset and cross-dataset evaluations, the latter of which considers the UTKFace dataset. From the experiments and analysis in the same-dataset evaluation, the highest validation accuracy of happens for the image of size pixels with eight patches. In comparison, the highest testing accuracy of occurs for the image of size pixels with patches. Moreover, the experiments and analysis in the cross-dataset evaluation show that the model works optimally for the image size pixels with patches, with the value of the model’s accuracy, precision, recall, and F1-score being , , , and , respectively. Furthermore, the misclassification analysis shows that the model works optimally in classifying the gender of people between 21-70 years old. The findings of this study can serve as a baseline for conducting further analysis on the effectiveness of gender classifier models considering various physical factors.
... This approach necessitates the use of specialized equipment such as fingerprint scanners, palm scanners, and DNA analysers to collect data for a large number of individuals. In the majority of biometric applications and targets, the device must be physically touched in order to obtain data (Liew, Hani, Radzi, & Bakhteri, 2016;Anggara et al., 2023;Ateş and Khameneh, 2023). Like biometrics, this approach allows for the differentiation of an individual's physical characteristics and is extensively utilized in security settings. ...
In the contemporary era, technology has become an indispensable component of our daily lives, augmenting our existence in manifold ways. Notably, face recognition and Internet of Things (IoT) have undergone significant advancements in performance over the last few years. This phenomenon can be attributed to two factors namely, optimization of cost and time in the classroom, while simultaneously ensuring accuracy in attendance, and the ability to connect with electrical equipment automatically. The current study proposes an automatic smart classroom application that utilizes the concepts of IoT and face recognition. Specifically, the application relies entirely on face detection and recognition to automatically detect the presence of students in the classroom and mark their attendance. Moreover, the application also provides advanced technology that enables automatic switching on of electrical power, including but not limited to lamps, fans, LCDs, and air conditioning systems. This feature not only reduces energy consumption but also allows for monitoring of units consumed. The proposed solution represents a significant step towards establishing a smart classroom. Results and analysis of the IoT solution demonstrate that it accelerates the availability process in the classroom without requiring human intervention.
... In [31], proposed an approach using CNN on SUMS and AT &T databases for real-time gender classification based on facial images. Here, unlike traditional CNNs, the convolution process has been replaced by cross-correlation. ...
Biometric authentication systems identify or verify a person from a digital image taken by security cameras or fingerprint readers. Digital images are used for authentication wherever a security system exists, such as in airports and banks. Although biometric data authentication boosts security, it has several practical challenges and is a difficult problem in computer vision. Another application classifies biometric data according to certain characteristics such as age, gender, or race. One of the biometric data frequently used for this purpose and has become very important is face images. Deep learning systems can learn rich, compact representations of faces from very big face datasets, allowing people to surpass their facial analysis talents. The Convolutional Neural Network (CNN) has recently obtained very promising face analysis results among these methods. Although CNN has the beneficial use of the data’s correlation information, it has trouble learning efficiently when the supplied amount of the data or model is too huge. Quantum Convolutional Neural Network (QCNN) provides a new solution to a CNN-related problem using a quantum computing environment. In this study, gender recognition is performed with CNN and QCNN algorithms, and the results are compared in terms of time and accuracy. The purpose of the study is to show the comparative evaluation of QCNN and its classical counterpart CNN algorithms with detailed applications under the same conditions. 92% accuracy for QCNN and 90% accuracy for CNN are obtained. The total processing time is 128.85 s for QCNN and 832.30 s for CNN.
... Liew et al. [18], an approach using CNNs is proposed for real-time gender classifications based on the image of the face. Liew et al. deliver the goods classification accuracy 98.75% and 99.38% on sums and AT&T databases, severally. ...
Stores sells consumer goods, mainly food products and other household products at retail. The products sold in stores vary greatly, in order to be time efficient in the fast-paced era and the current technological era requires artificial intelligence technology. In the artificial intelligence branch, there is a specific or detailed learning process known as deep learning. One of the branches of deep learning is the convolutional neural network(CNN). This research intends to employ a CNN architecture to facilitate and streamline the time and cost of the store’s product sorting process. The test is conducted with 1,050 product images divided into 35 labels and divided into three data, namely 80% data training 10% data validation and 10% data test. The image used is preprocessed with a size of 256×256 pixels. The data was trained with six convolution layers and an epoch of 50 with an execution time of 33 minutes so as to achieve an accuracy of 91.37%.
... They have compared performance of proposed system with other modern techniques. In [14], for real time gender classification they have used convolutional neural network (CNN). They have tried to classify gender with reduce complexity of CNN model. ...
... Different modalities (e.g., the face, iris, and periocular regions) have been employed to solve the GC problem. Some methods have relied on face images acquired through careful preparation and posing in a constrained environment; thus, these methods usually achieve good accuracy [2]. However, their performances decline significantly when photos feature real scenarios in which faces are presented under different conditions [3], such as the occlusions shown in Fig. 1. ...
... Sedangkan lapisan jaringan syaraf biasa di dalam CNN bertugas melakukan klasifikasi. Sebelum CNN banyak digunakan, pengenalan wajah dilakukan melalui tahapan-tahapan seperti ekstraksi fitur, memperkecil dimensi citra, dan klasifikasi dilakukan secara terpisah [16]. Hal ini menyebabkan kesesuaian algoritma yang digunakan pada setiap tahapan sangat diperlukan. ...
... Untuk klasifikasi jenis kelamin, CNN memiliki kinerja yang sangat baik dengan mendapatkan akurasi sebesar 97% [18]. Dengan arsitektur yang berbeda, CNN dapat memperoleh akurasi hingga 98,75% [16]. Untuk klasifikasi rentang usia berdasarkan wajah seseorang, dengan multi-class classification yang terdiri dari 8 kelas, CNN memperoleh akurasi 59,90% [19]. ...
Kepuasan pelanggan menunjukan seberapa baik produk atau layanan suatu organisasi secara keseluruhan dalam memenuhi harapan pelanggan. Ekspresi wajah pelanggan dapat menunjukan kepuasan mereka terhadap layanan yang diberikan. Convolution Neural Network (CNN) adalah jenis algoritma neural network yang dapat digunakan untuk mengenali suatu objek di dalam sebuah gambar. CNN memanfaatkan proses konvolusi untuk menentukan dan membedakan suatu objek di dalam gambar dengan objek lainnya seperti untuk mengenali berbagai ekspresi wajah. Penelitian ini bertujuan untuk mengukur kepuasan pelanggan dengan memanfaatkan model CNN dengan mengenali setiap perubahan ekspresi wajah. Dari hasil pelatihan model CNN diperoleh akurasi sebesar 90,57%. Selanjutnya model yang terbentuk diimplementasikan ke dalam sebuah sistem berbasis web yang melakukan perekaman ekspresi wajah dan melakukan klasifikasi (puas atau tidak puas) terhadap setiap perubahan wajah yang terdeteksi. Ekspresi yang paling dominan merupakan hasil dari pengukuran kepuasan pelanggan.
... C loss can be differently designed like (5), (7). M > 0 is a hyperparameter used to guide the optimization. ...
Siamese network has obtained growing attention in real-life applications. In this survey, we present an comprehensive review on Siamese network from the aspects of methodologies, applications and interesting topics for further exploration. We firstly introduce framework designs of Siamese network, followed by methodologies about learning with unlabeled data. Then, we review application scenarios in terms of classification and regression, together with relative methodologies. We also discuss the promising area of few-shot learning (FSL), followed by interesting topics about opportunities and challenges.
