Face expression recognition based on improved convolutional neural network

Background/Objectives: There is only limited research work is going on in the field of facial expression recognition on low resolution images. Mostly, all the images in the real world will be in low resolution and might also contain noise, so this study is to design a novel convolutional neural network model (FERConvNet), which can perform better on low resolution images. Methods: We proposed a model and then compared with state-of-art models on FER2013 dataset. There is no publicly available dataset, which contains low resolution images for facial expression recognition (Anger, Sad, Disgust, Happy, Surprise, Neutral, Fear), so we created a Low Resolution Facial Expression (LRFE) dataset, which contains more than 6000 images of seven types of facial expressions. The existing FER2013 dataset and LRFE dataset were used. These datasets were divided in the ratio 80:20 for training and testing and validation purpose. A HDM is proposed, which is a combination of Gaussian Filter, Bilateral Filter and Non local means denoising Filter. This hybrid denoising method helps us to increase the performance of the convolutional neural network. The proposed model was then compared with VGG16 and VGG19 models. Findings: The experimental results show that the proposed FERConvNet_HDM approach is effective than VGG16 and VGG19 in facial expression recognition on both FER2013 and LRFE dataset. The proposed FERConvNet_HDM approach achieved 85% accuracy on Fer2013 dataset, outperforming the VGG16 and VGG19 models, whose accuracies are 60% and 53% on Fer2013 dataset respectively. The same FERConvNet_HDM approach when applied on LRFE dataset achieved 95% accuracy. After analyzing the results, our FERConvNet_HDM approach performs better than VGG16 and VGG19 on both Fer2013 and LRFE dataset. Novelty/Applications: HDM with convolutional neural networks, helps in increasing the performance of convolutional neural networks in Facial expression recognition. Keywords: Facial expression recognition; facial emotion; convolutional neural network; deep learning; computer vision

We proposed a facial motion tracking and expression recognition system based on video data. By a 3D deformable facial model, the online statistical model (OSM) and cylinder head model (CHM) were combined to track 3D facial motion in the framework of particle filtering. For facial expression recognition, a fast and efficient algorithm and a robust and precise algorithm were developed. With the first, facial animation and facial expression were retrieved sequentially. After that facial animation was obtained, facial expression was recognized by static facial expression knowledge learned from anatomical analysis. With the second, facial animation and facial expression were simultaneously retrieved to increase the reliability and robustness with noisy input data. Facial expression was recognized by fusing static and dynamic facial expression knowledge, the latter of which was learned by training a multi-class expressional Markov process using a video database. The experiments showed that facial motion tracking by OSM+CHM is more pose robust than that by OSM, and the facial expression score of the robust and precise algorithm is higher than those of other state-of-the-art facial expression recognition methods.

A novel database of children's spontaneous facial expressions (LIRIS-CSE)

Three convolutional neural network models for facial expression recognition in the wild

Fine-tuning of pretrained convolutional neural networks (CNNs) has been proven to be an effective strategy for remote sensing image scene classification, particularly when a limited number of labeled data sets are available for training purposes. However, such a fine-tuning process often needs that the input images are resized into a fixed size to generate input vectors of the size required by fully connected layers (FCLs) in the pretrained CNN model. Such a resizing process often discards key information in the scenes and thus deteriorates the classification performance. To address this issue, in this paper, we introduce a scale-free CNN (SF-CNN) for remote sensing scene classification. Specifically, the FCLs in the CNN model are first converted into convolutional layers, which not only allow the input images to be of arbitrary sizes but also retain the ability to extract discriminative features using a traditional sliding-window-based strategy. Then, a global average pooling (GAP) layer is added after the final convolutional layer so that input images of arbitrary size can be mapped to feature maps of uniform size. Finally, we utilize the resulting feature maps to create a new FCL that is fed to a softmax layer for final classification. Our experimental results conducted using several real data sets demonstrate the superiority of the proposed SF-CNN method over several well-known classification methods, including pretrained CNN-based ones.

Automated emotion recognition (AEE) plays a crucial role in numerous industries that depend on understanding human emotional responses, such as advertising, technology, and human-robot interaction, particularly within the Information Technology (IT) field. However, current systems can often come up short in comprehensively understanding an individual's emotions, as prior research has mainly focused on assessing facial expressions and categorizing them into seven primary emotions, including neutrality. In this study, we present several Deep Convolutional Neural Network (CNN) models designed specifically for the task of facial emotion recognition, utilizing the FER2013 and RAF datasets. The baseline CNN model is established through a trial-and-error method, and its results are compared with more complex deep learning techniques, including ResNet18, VGGNet16, VGGNet19, and Efficient Net-B0 models. Among these models, the VGGNet19 model achieved the best results with a test accuracy of 71.02% on the FER2013 dataset. In comparison, the ResNet18 model outperformed all other models with an 86.02% test accuracy on the RAF-DB dataset. These results underscore the potential for advancing automated emotion recognition through complex deep-learning techniques.

Development in Computer vision applications, most prominently in Artificial Intelligence (AI) over the past decade, has encouraged researchers to use computational algorithms. Machine Learning (ML) and Deep Learning (DL) algorithms are getting attention in the biomedical area to early diagnosis and prognosis of lethiferous diseases with higher accuracy using statistical analysis. Cervical cancer is one of the predominant gynecological infectious forms of cancer in females ecumenically, generally transpiring in less-developed nations. DL algorithms are more optimized and convenient due to the automatic feature extraction in Cervical lesion detection at an early phase. Here we analyze the convolutional neural network (CNN) models' performance using variation in channels depth comprises in the convolution layers to classify multi-class liquid-based cytology (LBC) Whole slide images (WSI). We propose three CNN models having two convolution layers with the number of channels (4,8), (8,16), and (32,64) and two pooling layers. CNN model with the highest number of channels in the convolution layers gives the leading performance with an accuracy of 96.89%, precision of 93.38%, the sensitivity of 93.75%, and F-score of 94.15%.

In this paper, we proposed a novel algorithm for Facial Expression Recognition (FER) which was based on fusion of gabor texture features and Local Phase Quantization (LPQ). Firstly, the LPQ feature and gabor texture feature were respectively extracted from every expression image. LPQ features are histograms of LPQ transform. Five scales and eight orientations of gabor wavelet filters are used to extract gabor texture features and adaboost algorithm is used to select gabor features. Then we obtain two expression recognition results on both expression features by Sparse Representation-based Classification (SRC) method. Finally, the final expression recognition was performed by fusion of residuals of two SRC algorithms. The experiment results on Japanese Female Facial Expression (JAFFE) database demonstrated that the new algorithm was better than the original two algorithms, and this algorithm had a much higher recognition rate than the traditional algorithm.

Developing a treatment strategy that effectively prolongs the lives of people with brain tumors requires an accurate diagnosis of the condition. Therefore, improving the preoperative classification of meningiomas is a priority. Machine learning (ML) has made great strides thanks to the development of convolutional neural networks (CNNs) and computer-aided tumor detection systems. The deep convolutional layers automatically extract important and dependable information from the input space, in contrast to more traditional neural network layers. One recent and promising advancement in this field is ML. Still, there is a dearth of studies being carried out in this area. Therefore, starting with the analysis of magnetic resonance images, we have suggested in this research work a tried-and-tested and methodical strategy for real-time meningioma diagnosis by image segmentation using a very deep transfer learning CNN model or DNN model (VGG-16) with CUDA. Since the VGGNet CNN model has a greater level of accuracy than other deep CNN models like AlexNet, GoogleNet, etc., we have chosen to employ it. The VGG network that we have constructed with very small convolutional filters consists of 13 convolutional layers and 3 fully connected layers. Our VGGNet model takes in an sMRI FLAIR image input. The VGG's convolutional layers leverage a minimal receptive field, i.e., 3 × 3, the smallest possible size that still captures up/down and left/right. Moreover, there are also 1 × 1 convolution filters acting as a linear transformation of the input. This is followed by a ReLU unit. The convolution stride is fixed at 1 pixel to keep the spatial resolution preserved after convolution. All the hidden layers in our VGG network also use ReLU. A dataset consisting of 264 3D FLAIR sMRI image segments from three different classes (meningioma, tuberculoma, and normal) was employed. The number of epochs in the Sequential Model was set to 10. The Keras layers that we used were Dense, Dropout, Flatten, Batch Normalization, and ReLU. According to the simulation findings, our suggested model successfully classified all of the data in the dataset used, with a 99.0% overall accuracy. The performance metrics of the implemented model and confusion matrix for tumor classification indicate the model's high accuracy in brain tumor classification. The good outcomes demonstrate the possibility of our suggested method as a useful diagnostic tool, promoting better understanding, a prognostic tool for clinical outcomes, and an efficient brain tumor treatment planning tool. It was demonstrated that several performance metrics we computed using the confusion matrix of the previously used model were very good. Consequently, we think that the approach we have suggested is an important way to identify brain tumors.

It is fascinating to hear about the research conducted on facial expression recognition algorithms and its significance in the era of intelligence, especially with the development of the internet and IoT. The use of deep learning method and neural network VGG to study facial expressions sounds promising. It is encouraging to know that the network structure was improved and pooled using Soft pool, thereby enhancing the extraction of facial expression features. The integration of an attention mechanism to improve the ability of the network to recognize facial expressions appears to be a valuable addition to the research. it is impressive that the recognition rate of the improved network on the face expression dataset RAF-DB reached 86.1%, exceeding that of the original VGG network on this data set. this outcome suggests that the changed network is effective for facial expression recognition. thank you for sharing this information.

Understanding the learning mechanism of convolutional neural networks in spectral analysis

In deep convolutional neural networks (DCNNs), model size and computation complexity are two important factors governing throughput and energy efficiency when deployed to hardware for inference. Recent works on compact DCNNs as well as pruning methods are effective, yet with drawbacks. For instance, more than half the size of all MobileNet models lies in their last two layers, mainly because compact separable convolution (CONV) layers are not applicable to their last fully connected (FC) layers. Also, in pruning methods, the compression is gained at the expense of irregularity in the DCNN architecture, which necessitates additional indexing memory to address nonzero weights, thereby increasing memory footprint, decompression delays, and energy consumption. In this article, we propose cyclic sparsely connected (CSC) architectures, with memory/computation complexity of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathcal {O}(N\log {}N)$ </tex-math></inline-formula> , where <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$N$ </tex-math></inline-formula> is the number of nodes/channels given a DCNN layer that, contrary to compact depthwise separable layers, can be used as an overlay for both FC and CONV layers of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\mathcal {O}(N^{2})$ </tex-math></inline-formula> . Also, contrary to pruning methods, CSC architectures are structurally sparse and require no indexing due to their cyclic nature. We show that both standard convolution and depthwise convolution layers are special cases of the CSC layers, whose mathematical function, along with FC layers, can be unified into one single formulation and whose hardware implementation can be carried out under one arithmetic logic component. We examine the efficacy of the CSC architectures for compression of LeNet, AlexNet, and MobileNet DCNNs with precision ranging from 2 to 32 bits. More specifically, we surge upon the compact 8-bit quantized 0.5 MobileNet V1 and show that by compressing its last two layers with CSC architectures, the model is compressed by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\sim 1.5\times $ </tex-math></inline-formula> with a size of only 873 kB and little accuracy loss. Finally, we design a configurable hardware that implements all types of DCNN layers including FC, CONV, depthwise, CSC-FC, and CSC-CONV indistinguishably within a unified pipeline. We implement the hardware on a tiny Xilinx field-programmable gate array (FPGA) for total on-chip processing of the compressed MobileNet that, compared to the related work, has the highest Inference/J while utilizing the smallest FPGA.

This is the first work to propose generation of the network model of a convolutional neural network (CNN) specifically for power-efficient implementation using the ReRAM technology. State-of-the-art in this area is based on implementation of an already fixed CNN model. It uses parallel crossbar structures to achieve a desired precision quantization of the edge weights using the limited precision provided by individual ReRAM devices. In contrast, in this work we keep the ReRAM crossbar structure in mind during model generation, and target altering a base CNN model such that the resulting implementation will be more power efficient. This is by means of eliminating the parallel crossbars as much as possible, thus reducing the number of required analog-to-digital and digital-to-analog converters which are dominant sources of power consumption. We propose four architectural techniques which are applicable to convolutional and fully-connected layers in a CNN and alter the network model to work with lower precision of the weights with negligible loss in accuracy. In our experiments, we achieve at least 50% power savings with almost no loss in accuracy for popular CNNs compared to ReRAM implementation of the base model.

In this paper, the performance of Facial Expression Recognition (FER) using Deep Convolutional Neural Network (DCNN) model is evaluated. The expressions include Angry, Disgust, Fear, Happy, Sad, Surprise, and Neutral. In addition to performance evaluation, the analysis on Interclass false positives are also discussed which helps to analyze the underlying challenges to improve the model. All classifiers give low performance on Fer2013 datasets. DCNN gives 54.46% accuracy on test and 89.52% on training set, whereas, in case of different kernels of Support Vector Machines (SVM), the highest accuracy is 45% using cubic kernel on training set. Experiments show that certain facial expressions have more false positives and few of them are very dominant. In case of Disgust expression, it has Angry as dominant false positive. Based on false positives analysis, binary classifiers can be trained to improve the accuracy of the expressions with dominant false positives. Experiments on Fer2013 dataset confirms that the accuracy of expression Disgust is improved by piping the binary classifiers, Disgust vs. Angry, to existing DCNN of multi-class.

Ground short circuit faults in current transmission lines are common in the power systems. In order to prevent the power system from aggravating the accident caused by short-circuit faults of transmission lines, a novel convolutional neural network (CNN) model is constructed to identify the short-circuit fault of the transmission line in the power system. The CNN model is mainly consisted of five convolutional layers, three max-pooling layers, one concatenate layer, one dropout layer, one fully connected layer, and a Softmax classifier. This method uses a fixed time window to intercept system short-circuit fault data, extracts the deep features of these data from the training samples, and then corresponds the extracted features to labels one-to-one. Finally, in PSCAD/EMTDC, the new England 10 machine 39 nodes are taken as an example to realize the simulation. The experimental results show that the CNN model can quickly and accurately identify the short-circuit fault types, and the optimal model accuracy rate reaches 99.95%. The results of this manuscript -have positive effect on enhancing the disaster prevention capability and the operation stability of transmission lines.