Small-scale Face Research Articles

Research on improved small scale face detection algorithm based on Yolov5

To address the challenges of dense scenarios with densely distributed small-scale faces, severe occlusions, and unclear features leading to inaccurate detection and high miss rates, we propose a lightweight small-scale face detection algorithm based on YOLOv5. The aim is to enhance the accuracy and precision of target detection. Firstly, we introduce the Convolutional Block Attention Module (CBAM) into the existing backbone network, obtaining more detailed features by comprehensively considering both spatial and channel dimensions. Next, in the Neck network, we embed involution to enhance channel information and weight distribution. Finally, a new feature fusion layer is added to improve the capture capability of feature information for smaller pixels and smaller targets in visible areas by integrating deep semantic information with shallow semantic information. The experimental results demonstrate that the improved model exhibits an increase in the average precision across all three subsets of the public WIDER FACE dataset, with improvements of 3.2%, 3.4%, and 2.6% respectively. The detection frame rate reaches 87 frames per second (FPS), significantly enhancing the detection performance of facial targets. This improvement meets the accuracy and real-time requirements for detecting small-scale facial targets in dense scenarios.

Coverless image steganography using morphed face recognition based on convolutional neural network

In recent years, information security has become a prime issue of worldwide concern. To improve the validity and proficiency of the image data hiding approach, a piece of state-of-the-art secret information hiding transmission scheme based on morphed face recognition is proposed. In our proposed data hiding approach, a group of morphed face images is produced from an arranged small-scale face image dataset. Then, a morphed face image which is encoded with a secret message is sent to the receiver. The receiver uses powerful and robust deep learning models to recover the secret message by recognizing the parents of the morphed face images. Furthermore, we design two novel Convolutional Neural Network (CNN) architectures (e.g. MFR-Net V1 and MFR-Net V2) to perform morphed face recognition and achieved the highest accuracy compared with existing networks. Additionally, the experimental results show that the proposed schema has higher retrieval capacity and accuracy and it provides better robustness.

Optimization of face recognition algorithm based on deep learning multi feature fusion driven by big data

Today, with the rapid development of science and technology, the era of big data has been proposed and triggered reforms in all walks of life. Face recognition is a biometric recognition method with the characteristics of non-contact, non mandatory, friendly and harmonious, which has a good application prospect in the fields of national security and social security. With the deepening of the research on face recognition, small-scale face recognition has achieved good recognition results, but in the era of big data, the existing small-scale face recognition methods have gradually failed to meet the social needs, and how to get a good face recognition effect in the era of big data has become a new research hotspot. Based on this, this paper aims to optimize the existing face recognition algorithm, study the face recognition method driven by big data, and propose a deep learning multi feature fusion face recognition algorithm driven by big data. First, for the problem that 2DPCA (Two-dimensional Principle Component Analysis) can well extract the global features of the face under large samples, but the local features of the face are difficult to process, this paper uses the LBP (Local Binary Pattern, LBP) algorithm to extract the texture features of the face, and the extracted texture features are integrated with the global features extracted by 2DPCA to multi-feature fusion, so that the fused features can take into account both global and local features, and have better recognition results. Then using the obtained fusion features as input, training in a convolutional neural network, and measuring the similarity based on the feature vectors of the sample set and the training set after the training, can realize multi-feature fusion face recognition. Through the analysis of simulation experiments, it is found that, compared with the use of global features or local features alone, the fusion features obtained by multi-feature fusion of global features extracted by 2DPCA and local features extracted by LBP algorithm have better recognition effect in the big data environment. After convolutional neural network trains and recognizes this feature, a high recognition accuracy rate is obtained, which can show that the face recognition method designed in this paper has good application potential in the era of big data. In the background of big data, the accuracy of face recognition can reach more than 90%, which can meet the needs of society well.

Small-Scale Face Detection Based on Improved R-FCN

Face detection is an important basic technique for face-related applications, such as face analysis, recognition, and reconstruction. Images in unconstrained scenes may contain many small-scale faces. The features that the detector can extract from small-scale faces are limited, which will cause missed detection and greatly reduce the precision of face detection. Therefore, this study proposes a novel method to detect small-scale faces based on region-based fully convolutional network (R-FCN). First, we propose a novel R-FCN framework with the ability of feature fusion and receptive field adaptation. Second, a bottom-up feature fusion branch is established to enrich the local information of high-layer features. Third, a receptive field adaptation block (RFAB) is proposed to ensure that the receptive field can be adaptively selected to strengthen the expression ability of features. Finally, we improve the anchor setting method and adopt soft non-maximum suppression (SoftNMS) as the selection method of candidate boxes. Experimental results show that average precision for small-scale face detection of R-FCN with feature fusion branch and RFAB (RFAB-f-R-FCN) is improved by 0.8%, 2.9%, and 11% on three subsets of Wider Face compared with that of R-FCN.

High-Precision Portrait Classification Based on MTCNN and Its Application on Similarity Judgement

Portrait classification is a complex course including at least face detection, recognization and compare each of which contains multi-tasks, facing plenty of various challenging questions due to askew poses, illuminations, occlusions, image blurring and small scale face in the pictures. Though deep learning methods, such as Convolutional Neural Network (CNN) family and You Only Look Once (YOLO) series, had boomed a large number of areas on object detection and accelerated the solving of these difficulties on image processing, they are not specially designed for the image classification and may require a great deal of resource, expensive computation and taxing annotation. In 2016, an innovative face detection model named Multi-task convolutional neural network (MTCNN) arose and triggered viral and wide spread. Its high efficient and accurate performance on both face detection and face alignment tasks, real time effect based on lightweight CNN as well as effective conducting online hard sample mining, all contribute to significant improvement to the challenges above. This paper introduces the MTCNN algorithm and applies it to the similarity judgement with two industrial real problems together with FaceNet model. In addition, some effective practical methods on increase precision of classification are also proposed to gain better effect.

Real-time dissolved carbon dioxide monitoring I: Application of a novel in situ sensor for CO2 monitoring and control.

Dissolved carbon dioxide (dCO2) is a well‐known critical parameter in bioprocesses due to its significant impact on cell metabolism and on product quality attributes. Processes run at small‐scale faces many challenges due to limited options for modular sensors for online monitoring and control. Traditional sensors are bulky, costly, and invasive in nature and do not fit in small‐scale systems. In this study, we present the implementation of a novel, rate‐based technique for real‐time monitoring of dCO2 in bioprocesses. A silicone sampling probe that allows the diffusion of CO2 through its wall was inserted inside a shake flask/bioreactor and then flushed with air to remove the CO2 that had diffused into the probe from the culture broth (sensor was calibrated using air as zero‐point calibration). The gas inside the probe was then allowed to recirculate through gas‐impermeable tubing to a CO2 monitor. We have shown that by measuring the initial diffusion rate of CO2 into the sampling probe we were able to determine the partial pressure of the dCO2 in the culture. This technique can be readily automated, and measurements can be made in minutes. Demonstration experiments conducted with baker's yeast and Yarrowia lipolytica yeast cells in both shake flasks and mini bioreactors showed that it can monitor dCO2 in real‐time. Using the proposed sensor, we successfully implemented a dCO2‐based control scheme, which resulted in significant improvement in process performance.

SE-IYOLOV3: An Accurate Small Scale Face Detector for Outdoor Security

Small scale face detection is a very difficult problem. In order to achieve a higher detection accuracy, we propose a novel method, termed SE-IYOLOV3, for small scale face in this work. In SE-IYOLOV3, we improve the YOLOV3 first, in which the anchorage box with a higher average intersection ratio is obtained by combining niche technology on the basis of the k-means algorithm. An upsampling scale is added to form a face network structure that is suitable for detecting dense small scale faces. The number of prediction boxes is five times more than the YOLOV3 network. To further improve the detection performance, we adopt the SENet structure to enhance the global receptive field of the network. The experimental results on the WIDERFACEdataset show that the IYOLOV3 network embedded in the SENet structure can significantly improve the detection accuracy of dense small scale faces.

LiveNet: Improving features generalization for face liveness detection using convolution neural networks

Performance of face liveness detection algorithms in cross-database face liveness detection tests is one of the key issues in face-biometric based systems. Recently, Convolution Neural Networks (CNN) classifiers have shown remarkable performance in intra-database face liveness detection tests. However, a little effort has been made to improve the generalization capability of CNN classifiers for cross-database and unconstrained face liveness detection tests. In this paper, we propose an efficient strategy for training deep CNN classifiers for face liveness detection task. We utilize continuous data-randomization (like bootstrapping) in the form of small mini-batches during training CNN classifiers on small scale face anti-spoofing database. Experimental results revealed that the proposed approach reduces the training time by 18.39%, while significantly lowering the HTER by 8.28% and 14.14% in cross-database tests on CASIA-FASD and Replay-Attack database respectively as compared to state-of-the-art approaches. Additionally, the proposed approach achieves satisfactory results on intra-database and cross-database face liveness detection tests, claiming a good generality over other state-of-the-art face anti-spoofing approaches.