Feature extraction and classification methods of facial expression: a survey

Htay, Moe Moe

doi:10.11591/csit.v2i1.p26-32

Cited by 7 publications

(5 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…CNN is known as one of the deep learning techniques used for CV tasks. Specifically, CNN is developed from multilayer perceptron (MP) to process two-dimensional data such as images [7], [12], [13]. CNN technique has three layers which are divided into two main parts, feature learning, and classifier parts.…”

Section: Methods 21 State-of-the-art Techniquesmentioning

confidence: 99%

“…The transfer learning approach has been chosen for the approach of this research because the technique has utilized best practices for state-of-the-art models [5]- [7]. Particularly, the trained models for detecting building shapes from given images employ convolutional neural networks (CNN) architectures such as AlexNet [8], visual geometry group (VGG) [9], and ResNet [10].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Masking preprocessing in transfer learning for damage building detection

Toba

Bunyamin

Widyaya

et al. 2023

IJ-AI

View full text Add to dashboard Cite

The sudden climate change occurring in different places in the world has made disasters more unpredictable than before. In addition, responses are often late due to manual processes that have to be performed by experts. Consequently, major advances in computer vision (CV) have prompted researchers to develop smart models to help these experts. We need a strong image representation model, but at the same time, we also need to prepare for a deep learning environment at a low cost. This research attempts to develop transfer learning models using low-cost masking pre-processing in the experimental building damage (xBD) dataset, a large-scale dataset for advancing building damage assessment. The dataset includes eight types of disasters located in fifteen different countries and spans thousands of square kilometers of satellite images. The models are based on U-Net, i.e., AlexNet, visual geometry group (VGG)-16, and ResNet-34. Our experiments show that ResNet-34 is the best with an F1 score of 71.93%, and an intersection over union (IoU) of 66.72%. The models are built on a resolution of 1,024 pixels and use only first-tier images compared to the state-of-the-art baseline. For future orientations, we believe that the approach we propose could be beneficial to improve the efficiency of deep learning training.

show abstract

Section: Methods 21 State-of-the-art Techniquesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Masking preprocessing in transfer learning for damage building detection

Toba

Bunyamin

Widyaya

et al. 2023

IJ-AI

View full text Add to dashboard Cite

show abstract

“…There are a variety of prevalent techniques when it comes to methods based on appearance characteristics. Table 1 emphasizes the significance of the main distinctions between various feature extraction techniques in FER: [18]. The extraction of geometric features requires the accurate selection of facial feature points, which can be challenging in cases of low image quality or complex backgrounds.…”

Section: A Appearance Feature-based Methodsmentioning

confidence: 99%

Feature Extraction Techniques for Facial Expression Recognition (FER)

2023

IJSER

View full text Add to dashboard Cite

Facial expression recognition (FER) is a significant area of study in computer vision and affective computing. In numerous applications, such as human-computer interaction, emotion detection, and behavior analysis. Feature extraction is a crucial stage in facial expression recognition systems, as it involves extracting pertinent information from facial images in order to accurately represent various facial expressions. The purpose of this paper is to investigate and compare the various feature extraction techniques used in facial expression recognition, as well as their merits, limitations, and impact on overall system performance. Using benchmark datasets and performance metrics, the evaluation provides insight into the efficacy of various feature extraction methods. In this study, we propose a method for facial expression recognition that integrates deep learning, Principal Component Analysis (PCA), and Gray Level Co-occurrence Matrix (GLCM). Use 1D-CNN for classification.

show abstract

“…(4) The Field of Brain Science Research . The research of the BCI system originally originated from the study of brain cognitive science, such as the use of brain electricity to detect polygraphs, the study of mental activity based on brain electricity, and this article involved in driving fatigue detection based on BCI system [ 9 ]. The research of the BCI system is based on the research of brain science, and the research of the BCI system in turn provides help to the research of brain cognitive science.…”

Section: Eeg Signals and The Classification Of Driver Fatiguementioning

confidence: 99%

Classifying Driving Fatigue by Using EEG Signals

Zeng

Wang

2022

Computational Intelligence and Neuroscience

View full text Add to dashboard Cite

Fatigue driving is one of the main reasons for the occurrence of traffic accidents. Brain-computer interface, as a human-computer interaction method based on EEG signals, can communicate with the outside world and move freely through brain signals without relying on the peripheral neuromuscular system. In this paper, a simulation driving platform composed of driving simulation equipment and driving simulation software is used to simulate the real driving process. The EEG signals of the subjects are collected through simulated driving, and the EEG of five subjects is selected as the training sample, and the remaining one is the subject. As a test sample, perform feature extraction and classification experiments, select any set of normal signals and fatigue signals recorded in the driving fatigue experiment for data analysis, and then study the classification of driver fatigue levels. Experiments have proved that the PSO-H-ELM algorithm has only about 4% advantage compared with the average accuracy of the KNN algorithm and the SVM algorithm. The gap is not as big as expected, but as a new algorithm, it is applied to the detection of fatigue EEG. The two traditional algorithms are indeed more suitable. It shows that the driver fatigue level can be judged by detecting EEG, which will provide a basis for the development of on-board, real-time driving fatigue alarm devices. It will lay the foundation for traffic management departments to intervene in driving fatigue reasonably and provide a reliable basis for minimizing traffic accidents.

show abstract

Feature extraction and classification methods of facial expression: a survey

Cited by 7 publications

References 24 publications

Masking preprocessing in transfer learning for damage building detection

Masking preprocessing in transfer learning for damage building detection

Feature Extraction Techniques for Facial Expression Recognition (FER)

Classifying Driving Fatigue by Using EEG Signals

Contact Info

Product

Resources

About