Multiscale parallel deep CNN (mpdCNN) architecture for the real low-resolution face recognition for surveillance

Mishra, Nayaneesh Kumar; Dutta, Mainak; Singh, Satish Kumar

doi:10.1016/j.imavis.2021.104290

Cited by 25 publications

(9 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Mishra et al [ 39 ] introduced a multiscale parallel deep CNN to solve problems in low- and high-resolution images. Nadeem et al [ 40 ] proposed integrating frontal and profile face image recognition using different CNNs in parallel, combining their predictions based on a single voting scheme.…”

Section: Related Workmentioning

confidence: 99%

Analysis of Real-Time Face-Verification Methods for Surveillance Applications

et al. 2023

View full text Add to dashboard Cite

In the last decade, face-recognition and -verification methods based on deep learning have increasingly used deeper and more complex architectures to obtain state-of-the-art (SOTA) accuracy. Hence, these architectures are limited to powerful devices that can handle heavy computational resources. Conversely, lightweight and efficient methods have recently been proposed to achieve real-time performance on limited devices and embedded systems. However, real-time face-verification methods struggle with problems usually solved by their heavy counterparts—for example, illumination changes, occlusions, face rotation, and distance to the subject. These challenges are strongly related to surveillance applications that deal with low-resolution face images under unconstrained conditions. Therefore, this paper compares three SOTA real-time face-verification methods for coping with specific problems in surveillance applications. To this end, we created an evaluation subset from two available datasets consisting of 3000 face images presenting face rotation and low-resolution problems. We defined five groups of face rotation with five levels of resolutions that can appear in common surveillance scenarios. With our evaluation subset, we methodically evaluated the face-verification accuracy of MobileFaceNet, EfficientNet-B0, and GhostNet. Furthermore, we also evaluated them with conventional datasets, such as Cross-Pose LFW and QMUL-SurvFace. When examining the experimental results of the three mentioned datasets, we found that EfficientNet-B0 could deal with both surveillance problems, but MobileFaceNet was better at handling extreme face rotation over 80 degrees.

show abstract

Section: Related Workmentioning

confidence: 99%

Analysis of Real-Time Face-Verification Methods for Surveillance Applications

et al. 2023

View full text Add to dashboard Cite

show abstract

“…In contrast, traditional neural networks are limited for classification task and require well-defined feature (engineered features) to achieve high performance [11], [12]. In general, CNNs are widely applied in many computer vision applications such as face recognition [13], object detection [14], Natural language processing [15], medical image analysis and others [16]. Fig.…”

Section: B Convolutional Neural Networkmentioning

confidence: 99%

Classification of Palm Trees Diseases using Convolution Neural Network

Abu-zanona¹,

Elaiwat²,

Younis³

et al. 2022

IJACSA

View full text Add to dashboard Cite

The palm tree is considered one of the most durable trees , and it occupies an advanced position as one of the most famous and most important trees that are planted in different regions around the world, which enter into many uses and have a number of benefits. In the recent years , date palms have been exposed to a large number of diseases. These diseases differ in their symptoms and causes, and sometimes overlap, making the diagnosing process with the naked eye difficult, even by an expert in this field. This paper proposes a CNN-model to detect and classify four common diseases threatening palms today, Bacterial leaf blight, Brown spots, Leaf smut, white scale in addition to healthy leaves. The proposed CNN structure includes four convolutional layers for feature extraction followed by a fully connected layer for classification. For performance evaluation, we investigate the performance of the proposed model and compare to other CNN-structures, VGG-16 and MobileNet, using four evaluation metrics: Accuracy, Precision, Recall and F1 Score. Our proposed model achieves 99.10% accuracy rate while VGG-16 and MobileNet achieve 99.35% and 99.56% accuracy rates, respectively. In general, the performance of our model and other models are very close with a minor advantage to MobileNet over others. In contrast, our model characterized by simplicity and shows low computational training time comparing to others.

show abstract

“…Also for applications like crime detection, emotion analysis would be helpful. Mishra et al (2021) propose the use of multiscale parallel deep CNN (mpdCNN) Architecture for face recognition of low resolution images for monitoring purpose [3]. The accuracy of the proposed architecture is 88.6% on the SCface database for low resolution images which is a good improvement than other architectures being used.…”

Section: Literature Surveymentioning

confidence: 99%

Deep Learning Based Hybrid Approach For Facial Emotion Detection

Sundaramoorthy¹,

Babuvignesh²

2021

Proceedings of the First International Conference on Combinatorial and Optimization, ICCAP 2021, December 7-8 2021, Chennai, In

View full text Add to dashboard Cite

Humans exhibit an essential set of feelings which might be exhibited via steady facial expressions. Emotions have an important role in teaching as well as other disciplines. It is understood that, while culture, the environment, the language and behavior of people vary, emotions are considered to be universal among the entire human population. Emotions are also important in the realm of education, because their power rivals that of the environment or the language used to communicate information. It's important for cognitive functions including learning and digesting new information.Student emotions during lectures play an important role whether it's in classrooms or in virtual learning environments. Unfortunately, due to the adoption of a virtual method of study or simply the incapacity of humans to keep track of all students in a classroom, teachers do not always keep students' emotions in check, making it difficult to adjust and keep the line of communication open.Thus automated facial emotion detection is very useful and brings the hidden indicators of students internal emotions to limelight. The suggested method aids in the detection of emotions and their classification in order to detect interest in a topic, which is then used as feedback to the course management staff in order to improve learner experience and aid in course material updates.

show abstract

Multiscale parallel deep CNN (mpdCNN) architecture for the real low-resolution face recognition for surveillance

Cited by 25 publications

References 18 publications

Analysis of Real-Time Face-Verification Methods for Surveillance Applications

Analysis of Real-Time Face-Verification Methods for Surveillance Applications

Classification of Palm Trees Diseases using Convolution Neural Network

Deep Learning Based Hybrid Approach For Facial Emotion Detection

Contact Info

Product

Resources

About