Facial Expression Recognition from a Single Face Image Based on Deep Learning and Broad Learning

Bie, Mei; Xu, Huan; Gao, Yan; Che, Xiangjiu

doi:10.1155/2022/7094539

Cited by 9 publications

(3 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The single-stage approach has the advantages of decreased memory utilisation, a simpler inference procedure, and faster computing speed in feature extraction. [6] The six primary emotions that are represented in human facial expressions are happiness, sadness, anger, fear, contempt, and surprise. Pre-processing face photos, extracting expression traits, and recognising facial expressions are the three primary components of facial expression recognition (FER).…”

Section: Experimental Methods or Methodologymentioning

confidence: 99%

Deep Learning-Based Small Face Detection from Hard Image

Sapna Shinde,

Priti Chakurkar,

Rashmi Rane

2024

Int Res J Adv Engg Hub

View full text Add to dashboard Cite

Facial detection usually comes first in face recognition and face analysis systems. Previously, techniques such as directed gradient histograms and cascades relied on manually-engineered features from particular photos. Nevertheless, the precision with which these techniques could identify faces in uncontrolled environments was restricted. Numerous deep learning-based face recognition frameworks have recently been developed, many of which have significantly increased accuracy, as a result of the rapid progress of deep learning in computer vision. Despite these advancements, detecting small, scaled, positioned, occluded, blurred, and faces that are partially occluded in uncontrolled conditions remains a challenge in face identification. This problem has been studied for many years but has not been completely resolved.

show abstract

Section: Experimental Methods or Methodologymentioning

confidence: 99%

Deep Learning-Based Small Face Detection from Hard Image

Sapna Shinde,

Priti Chakurkar,

Rashmi Rane

2024

Int Res J Adv Engg Hub

View full text Add to dashboard Cite

show abstract

“…DBN have many advantages, such as the ability to encode deeper, higher-order network topological structures, and the ability to prevent over-fitting and falling into local minima through special, inappropriate pre-training phases [26], the ability of unsupervised learning, fast inference, and multilayer structures [24]. Currently, DBN has already been applied to text detection [27][28][29], face and expression recognition [30,31], and hyperspectral image classification [32,33]. DBN has also been combined with AE.…”

Section: Introductionmentioning

confidence: 99%

An acoustic emission identification model for train axle fatigue cracks based on deep belief network

Lin,

Tang,

Zhu

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

Aiming to effectively identify train axle fatigue cracks, some scholars are now trying to consider introducing acoustic emission detection technology into axle health monitoring and combining it with intelligent neural networks to achieve better monitoring results. In the field of axle fatigue crack acoustic emission identification, the commonly used methods include parameter analysis, wavelet analysis, and traditional artificial neural networks. Though these methods did work in the field of axle fatigue crack identification, they still have two deficiencies. （1）The features are manually extracted depending on large artificial experience. (2) The methods mentioned above have shallow architectures, which cannot handle some complex fatigue crack linear relationships. In this paper, an acoustic emission signal analysis method for axle fatigue cracks based on deep belief networks is proposed to overcome the aforementioned deficiencies. The main purpose of this paper is to emphasize the feasibility of the deep belief network to intelligently identify fatigue crack signals from different types of acoustic emission signals of axles and to classify crack signals at different stages. The research method is to use acoustic emission (AE) technology to obtain the axle fatigue crack AE signal, pretreat the collected time-domain signal, identify whether cracks appear, and classify the stages of crack emergence by deep belief network (DBN) so as to find the axle crack in advance. The experimental analysis results show that the method of identification of axle fatigue cracks based on DBN not only can identify fatigue crack signals but also has a very high identification rate of fatigue cracks at different stages. In the axle fatigue crack acoustic emission identification field, it can be seen that the proposed method in this paper will be a promising approach.

show abstract

“…ML also help feature extraction from image data associated to emotional reaction. Notable works include the study by (Bie et al, 2022), which employed deep neural networks to extract facial expression features from images. Also, use a multilevel hierarchy deep learning approach to extract features from facial images and classify them into one of seven emotion categories (happiness, sadness, anger, surprise, disgust, fear, and neutral) (Savchenko, 2021).…”

Section: Machine Learning For Analysing Emotional Responsesmentioning

confidence: 99%

Decoding the Pattern Between Architecture and Emotion

Maghool

View full text Add to dashboard Cite

In this research, I investigated the interrelation between emotion and visual experience of buildings, specifically from an architect's perspective. To address this question, I extensively reviewed literature from various fields including architecture, psychology, and computer science. Based on this comprehensive investigation, I developed a framework that allowed for a controlled and scientific study of how buildings influence emotions. For testing this framework, I selected two architecturally significant "sacred" buildings: the Sheikh Lotfollah Mosque in Isfahan, Iran, and the Futuna Chapel in Wellington, New Zealand.As part of the framework, I created digital twins of the selected buildings using photogrammetry technology, enabling realistic virtual reality (VR) simulations. The high fidelity of these VR simulations greatly contributed to the framework's development and enhanced the ecological validity of the investigation. To gauge low-level emotional responses, I collected physiological data through a biosensor, specifically Electrodermal Activities (EDA), from 12 human subjects. The collected data underwent a series of processing techniques to ensure interpretability and suitability for further analysis.Moreover, I developed nine computational algorithms based on the Processing Fluency Theory to quantify users' visual experience during virtual visits to the buildings. These algorithms measured various aspects such as Light Contrast, Visual Complexity, Visual Fractality, Visual Symmetry, and other qualities, providing numerical indicators of the perception of these architectural features. In the initial phase, I conducted standard statistical comparisons between the visual experiences of the two buildings through data analysis. The findings of this phase demonstrated the algorithms' ability to describe architectural qualities using quantitative measures.In the subsequent phase, I employed a Machine Learning approach for data analysis. By training Machine Learning models, I aimed to uncover complex patterns between the visual features extracted from participants' architectural experiences and the emotional indicators obtained from the biosensors. The results revealed promising potential for the developed methodology. Consequently, this research suggests that the framework can be further utilized for systematic and scientifically grounded investigations into how the visual experience of architecture influences users' emotions. Overall, this study contributes to the understanding of the intricate relationship between emotion and the visual aspects of buildings, providing valuable insights for architects and researchers interested in designing spaces that elicit specific emotional responses.

show abstract

Facial Expression Recognition from a Single Face Image Based on Deep Learning and Broad Learning

Cited by 9 publications

References 43 publications

Deep Learning-Based Small Face Detection from Hard Image

Deep Learning-Based Small Face Detection from Hard Image

An acoustic emission identification model for train axle fatigue cracks based on deep belief network

Decoding the Pattern Between Architecture and Emotion

Contact Info

Product

Resources

About