Core point is a location that exhibits high curvature properties in a fingerprint. Detecting the accurate location of a core point is useful for efficient fingerprint matching, classification, and identification tasks. This paper proposes CP-Net, a novel core point detection network that comprises the Macro Localization Network (MLN) and the Micro-Regression Network (MRN). MLN is a specialized autoencoder network with an hourglass network at its bottleneck. It takes an input fingerprint image and outputs a region of interest that could be the most probable region containing the core point. The second component, MRN, regresses the RoI and locates the coordinates of the core point in the given fingerprint sample. Introducing an hourglass network in the MLN bottleneck ensures multi-scale spatial attention that captures local and global contexts and facilitates a higher localization accuracy for that area. Unlike existing multi-stage models, the components are stacked and trained in an end-to-end manner. Experiments have been performed on three widely used publicly available fingerprint datasets, namely, FVC2002 DB1A, FVC2004 DB1A, and FVC2006 DB2A. The proposed model achieved a true detection rate (TDR) of 98%, 100%, and 99.04% respectively, while considering 20 pixels distance from the ground truth as correct. Obtained experimental results on the considered datasets demonstrate that CP-Net outperforms the state-of-the-art core point detection techniques.
{"title":"CP-Net: Multi-Scale Core Point Localization in Fingerprints Using Hourglass Network","authors":"Geetika Arora, Arsh Kumbhat, Ashutosh Bhatia, Kamlesh Tiwari","doi":"10.1109/IWBF57495.2023.10157521","DOIUrl":"https://doi.org/10.1109/IWBF57495.2023.10157521","url":null,"abstract":"Core point is a location that exhibits high curvature properties in a fingerprint. Detecting the accurate location of a core point is useful for efficient fingerprint matching, classification, and identification tasks. This paper proposes CP-Net, a novel core point detection network that comprises the Macro Localization Network (MLN) and the Micro-Regression Network (MRN). MLN is a specialized autoencoder network with an hourglass network at its bottleneck. It takes an input fingerprint image and outputs a region of interest that could be the most probable region containing the core point. The second component, MRN, regresses the RoI and locates the coordinates of the core point in the given fingerprint sample. Introducing an hourglass network in the MLN bottleneck ensures multi-scale spatial attention that captures local and global contexts and facilitates a higher localization accuracy for that area. Unlike existing multi-stage models, the components are stacked and trained in an end-to-end manner. Experiments have been performed on three widely used publicly available fingerprint datasets, namely, FVC2002 DB1A, FVC2004 DB1A, and FVC2006 DB2A. The proposed model achieved a true detection rate (TDR) of 98%, 100%, and 99.04% respectively, while considering 20 pixels distance from the ground truth as correct. Obtained experimental results on the considered datasets demonstrate that CP-Net outperforms the state-of-the-art core point detection techniques.","PeriodicalId":273412,"journal":{"name":"2023 11th International Workshop on Biometrics and Forensics (IWBF)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123730504","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-19DOI: 10.1109/IWBF57495.2023.10157282
Marco Huber, Philipp Terhörst, Florian Kirchbuchner, Arjan Kuijper, N. Damer
Estimating and understanding uncertainty in face recognition systems is receiving increasing attention as face recognition systems spread worldwide and process privacy and security-related data. In this work, we investigate how such uncertainties can be further utilized to increase the accuracy and therefore the trust of automatic face recognition systems. We propose to use the uncertainties of extracted face features to compute a new uncertainty-aware comparison score (UACS). This score takes into account the estimated uncertainty during the calculation of the comparison score, leading to a reduction in verification errors. To achieve this, we model the comparison score and its uncertainty as a probability distribution and measure its distance to a distribution of an ideal genuine comparison. In extended experiments with three face recognition models and on six benchmarks, we investigated the impact of our approach and demonstrated its benefits in enhancing the verification performance and the genuine-imposter comparison scores separability.
{"title":"Uncertainty-aware Comparison Scores for Face Recognition","authors":"Marco Huber, Philipp Terhörst, Florian Kirchbuchner, Arjan Kuijper, N. Damer","doi":"10.1109/IWBF57495.2023.10157282","DOIUrl":"https://doi.org/10.1109/IWBF57495.2023.10157282","url":null,"abstract":"Estimating and understanding uncertainty in face recognition systems is receiving increasing attention as face recognition systems spread worldwide and process privacy and security-related data. In this work, we investigate how such uncertainties can be further utilized to increase the accuracy and therefore the trust of automatic face recognition systems. We propose to use the uncertainties of extracted face features to compute a new uncertainty-aware comparison score (UACS). This score takes into account the estimated uncertainty during the calculation of the comparison score, leading to a reduction in verification errors. To achieve this, we model the comparison score and its uncertainty as a probability distribution and measure its distance to a distribution of an ideal genuine comparison. In extended experiments with three face recognition models and on six benchmarks, we investigated the impact of our approach and demonstrated its benefits in enhancing the verification performance and the genuine-imposter comparison scores separability.","PeriodicalId":273412,"journal":{"name":"2023 11th International Workshop on Biometrics and Forensics (IWBF)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132021649","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-19DOI: 10.1109/IWBF57495.2023.10157796
Ziga Babnik, N. Damer, V. Štruc
Contemporary face recognition (FR) models achieve near-ideal recognition performance in constrained settings, yet do not fully translate the performance to unconstrained (real-world) scenarios. To help improve the performance and stability of FR systems in such unconstrained settings, face image quality assessment (FIQA) techniques try to infer sample-quality information from the input face images that can aid with the recognition process. While existing FIQA techniques are able to efficiently capture the differences between high and low quality images, they typically cannot fully distinguish between images of similar quality, leading to lower performance in many scenarios. To address this issue, we present in this paper a supervised quality-label optimization approach, aimed at improving the performance of existing FIQA techniques. The developed optimization procedure infuses additional information (computed with a selected FR model) into the initial quality scores generated with a given FIQA technique to produce better estimates of the “actual” image quality. We evaluate the proposed approach in comprehensive experiments with six state-of-the-art FIQA approaches (CR-FIQA, FaceQAN, SER-FIQ, PCNet, MagFace, SER-FIQ) on five commonly used benchmarks (LFW, CFP-FP, CPLFW, CALFW, XQLFW) using three targeted FR models (ArcFace, ElasticFace, CurricularFace) with highly encouraging results.
{"title":"Optimization-Based Improvement of Face Image Quality Assessment Techniques","authors":"Ziga Babnik, N. Damer, V. Štruc","doi":"10.1109/IWBF57495.2023.10157796","DOIUrl":"https://doi.org/10.1109/IWBF57495.2023.10157796","url":null,"abstract":"Contemporary face recognition (FR) models achieve near-ideal recognition performance in constrained settings, yet do not fully translate the performance to unconstrained (real-world) scenarios. To help improve the performance and stability of FR systems in such unconstrained settings, face image quality assessment (FIQA) techniques try to infer sample-quality information from the input face images that can aid with the recognition process. While existing FIQA techniques are able to efficiently capture the differences between high and low quality images, they typically cannot fully distinguish between images of similar quality, leading to lower performance in many scenarios. To address this issue, we present in this paper a supervised quality-label optimization approach, aimed at improving the performance of existing FIQA techniques. The developed optimization procedure infuses additional information (computed with a selected FR model) into the initial quality scores generated with a given FIQA technique to produce better estimates of the “actual” image quality. We evaluate the proposed approach in comprehensive experiments with six state-of-the-art FIQA approaches (CR-FIQA, FaceQAN, SER-FIQ, PCNet, MagFace, SER-FIQ) on five commonly used benchmarks (LFW, CFP-FP, CPLFW, CALFW, XQLFW) using three targeted FR models (ArcFace, ElasticFace, CurricularFace) with highly encouraging results.","PeriodicalId":273412,"journal":{"name":"2023 11th International Workshop on Biometrics and Forensics (IWBF)","volume":"25 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121948022","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-19DOI: 10.1109/IWBF57495.2023.10157651
A. Fathan, J. Alam
One of the most widely used self-supervised (SS) speaker verification (SV) system training methods is to optimize the speaker embedding network in a discriminative fashion using clustering algorithm (CA)-driven Pseudo-Labels (PLs). Although the PL-based SS training scheme showed impressive performance, recent studies have shown that label noise can significantly impact performance. In this paper, we have explored various PLs driven by different CAs and conducted a fine-grained analysis of the relationship between the quality of the PLs and the SV performance. Experimentally, we shed light on several previously overlooked aspects of the PLs that can impact SV performance. Moreover, we could observe that the SS-SV performance is heavily dependent on multiple qualitative aspects of the CA used to generate the PLs. Furthermore, we show that SV performance can be severely degraded from overfitting the noisy PLs and that the mixup strategy can mitigate the memorization effects of label noise.
{"title":"On the influence of the quality of pseudo-labels on the self-supervised speaker verification task: a thorough analysis","authors":"A. Fathan, J. Alam","doi":"10.1109/IWBF57495.2023.10157651","DOIUrl":"https://doi.org/10.1109/IWBF57495.2023.10157651","url":null,"abstract":"One of the most widely used self-supervised (SS) speaker verification (SV) system training methods is to optimize the speaker embedding network in a discriminative fashion using clustering algorithm (CA)-driven Pseudo-Labels (PLs). Although the PL-based SS training scheme showed impressive performance, recent studies have shown that label noise can significantly impact performance. In this paper, we have explored various PLs driven by different CAs and conducted a fine-grained analysis of the relationship between the quality of the PLs and the SV performance. Experimentally, we shed light on several previously overlooked aspects of the PLs that can impact SV performance. Moreover, we could observe that the SS-SV performance is heavily dependent on multiple qualitative aspects of the CA used to generate the PLs. Furthermore, we show that SV performance can be severely degraded from overfitting the noisy PLs and that the mixup strategy can mitigate the memorization effects of label noise.","PeriodicalId":273412,"journal":{"name":"2023 11th International Workshop on Biometrics and Forensics (IWBF)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127200154","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-19DOI: 10.1109/IWBF57495.2023.10157791
Andrea Macarulla Rodriguez, Z. Geradts, M. Worring, Luis Unzueta
The accuracy of face recognition in real-world surveillance videos plays a crucial role in forensic investigation and security monitoring systems. Despite advancements in technology, face recognition methods can be influenced by variations in pose, illumination, and facial expression that often occur in these videos. To address this issue, we propose a new method for images-to-video face recognition that pairs face images with multiple attributes (soft labels) and face image quality (FIQ). This is followed by the application of three calibration methods to estimate the likelihood ratio, which is a statistical measure commonly used in forensic investigations. To validate the results, we test our method on the ENFSI proficiency test 2015 dataset, using SCFace and ForenFace as calibration datasets and three embedding models: ArcFace, FaceNet, and QMagFace. Our results indicate that using only high quality frames can improve face recognition performance for forensic purposes compared to using all frames. The best results were achieved when using the highest number of common attributes between the reference image and selected frames, or by creating a single common embedding from the selected frames, weighted by the quality of each frame’s face image.
{"title":"Improved Likelihood Ratios for Surveillance Video Face Recognition with Multimodal Feature Pairing","authors":"Andrea Macarulla Rodriguez, Z. Geradts, M. Worring, Luis Unzueta","doi":"10.1109/IWBF57495.2023.10157791","DOIUrl":"https://doi.org/10.1109/IWBF57495.2023.10157791","url":null,"abstract":"The accuracy of face recognition in real-world surveillance videos plays a crucial role in forensic investigation and security monitoring systems. Despite advancements in technology, face recognition methods can be influenced by variations in pose, illumination, and facial expression that often occur in these videos. To address this issue, we propose a new method for images-to-video face recognition that pairs face images with multiple attributes (soft labels) and face image quality (FIQ). This is followed by the application of three calibration methods to estimate the likelihood ratio, which is a statistical measure commonly used in forensic investigations. To validate the results, we test our method on the ENFSI proficiency test 2015 dataset, using SCFace and ForenFace as calibration datasets and three embedding models: ArcFace, FaceNet, and QMagFace. Our results indicate that using only high quality frames can improve face recognition performance for forensic purposes compared to using all frames. The best results were achieved when using the highest number of common attributes between the reference image and selected frames, or by creating a single common embedding from the selected frames, weighted by the quality of each frame’s face image.","PeriodicalId":273412,"journal":{"name":"2023 11th International Workshop on Biometrics and Forensics (IWBF)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129119377","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-19DOI: 10.1109/IWBF57495.2023.10157166
Javier Galbally, A. Cepilovs, R. Blanco-Gonzalo, G. Ormiston, O. Miguel-Hurtado, I. S. Racz
Even though some initial works have shown on small sets of data that not all fingerprints present the same level of utility for recognition purposes, there is still insufficient data-supported evidence to understand the impact that finger type may have on fingerprint quality and, in turn, also on fingerprint comparison. The present work addresses this still under-researched topic, on a large-scale database of operational data containing 10-print impressions of over 18,000 subjects. The results show a noticeable difference in the quality level of fingerprints produced by each of the 10 fingers and also between the dominant and non-dominant hands. Based on these observations, several recommendations are made regarding: 1) the selection of fingers to be captured depending on the context of the application; 2) improvement in the usability of scanners and the capturing protocols; 3) improvement in the development, ergonomics and positioning of the acquisition devices; and 4) improvement of recognition algorithms by incorporating information on finger type and handedness.
{"title":"Fingerprint quality per individual finger type: A large-scale study on real operational data","authors":"Javier Galbally, A. Cepilovs, R. Blanco-Gonzalo, G. Ormiston, O. Miguel-Hurtado, I. S. Racz","doi":"10.1109/IWBF57495.2023.10157166","DOIUrl":"https://doi.org/10.1109/IWBF57495.2023.10157166","url":null,"abstract":"Even though some initial works have shown on small sets of data that not all fingerprints present the same level of utility for recognition purposes, there is still insufficient data-supported evidence to understand the impact that finger type may have on fingerprint quality and, in turn, also on fingerprint comparison. The present work addresses this still under-researched topic, on a large-scale database of operational data containing 10-print impressions of over 18,000 subjects. The results show a noticeable difference in the quality level of fingerprints produced by each of the 10 fingers and also between the dominant and non-dominant hands. Based on these observations, several recommendations are made regarding: 1) the selection of fingers to be captured depending on the context of the application; 2) improvement in the usability of scanners and the capturing protocols; 3) improvement in the development, ergonomics and positioning of the acquisition devices; and 4) improvement of recognition algorithms by incorporating information on finger type and handedness.","PeriodicalId":273412,"journal":{"name":"2023 11th International Workshop on Biometrics and Forensics (IWBF)","volume":"155 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123296506","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-19DOI: 10.1109/IWBF57495.2023.10157416
Kai Zeng, Xiangyu Yu, Beibei Liu, Yu Guan, Yongjian Hu
The adverse impact of deepfakes has recently raised world-wide concerns. Many ways of deepfake detection are published in the literature. The reported results of existing methods are generally good under known settings. However, the robustness challenge in deepfake detection is not well addressed. Most detectors fail to identify deepfakes that have undergone post-processing. Observing that the conventionally adopted RGB space does not guarantee the best performance, we propose other color spaces that prove to be more effective in detecting corrupted deepfake videos. We design a robust detection approach that leverages an adaptive manipulation trace extraction network to reveal artifacts from two color spaces. To mimic practical scenarios, we conduct experiments to detect images with post-processings that are not seen in the training stage. The results demonstrate that our approach outperforms state-of-the-art methods, boosting the average detection accuracy by 7% ~ 17%.
{"title":"Detecting Deepfakes in Alternative Color Spaces to Withstand Unseen Corruptions","authors":"Kai Zeng, Xiangyu Yu, Beibei Liu, Yu Guan, Yongjian Hu","doi":"10.1109/IWBF57495.2023.10157416","DOIUrl":"https://doi.org/10.1109/IWBF57495.2023.10157416","url":null,"abstract":"The adverse impact of deepfakes has recently raised world-wide concerns. Many ways of deepfake detection are published in the literature. The reported results of existing methods are generally good under known settings. However, the robustness challenge in deepfake detection is not well addressed. Most detectors fail to identify deepfakes that have undergone post-processing. Observing that the conventionally adopted RGB space does not guarantee the best performance, we propose other color spaces that prove to be more effective in detecting corrupted deepfake videos. We design a robust detection approach that leverages an adaptive manipulation trace extraction network to reveal artifacts from two color spaces. To mimic practical scenarios, we conduct experiments to detect images with post-processings that are not seen in the training stage. The results demonstrate that our approach outperforms state-of-the-art methods, boosting the average detection accuracy by 7% ~ 17%.","PeriodicalId":273412,"journal":{"name":"2023 11th International Workshop on Biometrics and Forensics (IWBF)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129436642","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-19DOI: 10.1109/IWBF57495.2023.10157760
J. Flusser, F. Šroubek, J. Kamenický, B. Zitová
Visual data, such as images and videos, are frequently used as evidence in court trials. If the data quality is insufficient to convince the court, a carefully tailored data processing algorithm supported with expert’s opinion is necessary. We present two real cases from our forensic expertise practice, in which we demonstrate a successful application of video superresolution that helped to convict offenders. The most important feature of image processing algorithms to be legally accepted by the court, is to rule out artifacts with realistic details, which are known to appear for example in deep learning methods.
{"title":"Video superresolution in real forensic cases","authors":"J. Flusser, F. Šroubek, J. Kamenický, B. Zitová","doi":"10.1109/IWBF57495.2023.10157760","DOIUrl":"https://doi.org/10.1109/IWBF57495.2023.10157760","url":null,"abstract":"Visual data, such as images and videos, are frequently used as evidence in court trials. If the data quality is insufficient to convince the court, a carefully tailored data processing algorithm supported with expert’s opinion is necessary. We present two real cases from our forensic expertise practice, in which we demonstrate a successful application of video superresolution that helped to convict offenders. The most important feature of image processing algorithms to be legally accepted by the court, is to rule out artifacts with realistic details, which are known to appear for example in deep learning methods.","PeriodicalId":273412,"journal":{"name":"2023 11th International Workshop on Biometrics and Forensics (IWBF)","volume":"164 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116293600","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-19DOI: 10.1109/IWBF57495.2023.10157482
Shuai Shao, Victor Sanchez
Human activity recognition (HAR) is a core research topic in mobile and wearable computing, and has been applied in many applications including biometrics, health monitoring and sports coaching. In recent years, researchers have focused more attention on sensor-based HAR due to the popularity of sensor devices. However, sensor-based HAR faces the challenge of limited data size caused by the high cost of data collection and labelling work, resulting in low performance for HAR tasks. Data transformation and generative adversarial network (GAN) have been proposed as data augmentation approaches to enrich sensor data, thereby addressing the problem of data size limitations. In this paper, we studied the effectiveness of diffusion-based generative models for generating synthetic sensor data as compared to the other data augmentation approaches in sensor-based HAR. In addition, UNet has been redesigned in order to improve the efficiency and practicality of diffusion modelling. Experiments on two public datasets showed the performance of diffusion modelling compared with different data augmentation methods, indicating the feasibility of synthetic sensor data generated using diffusion modelling.
{"title":"A Study on Diffusion Modelling For Sensor-based Human Activity Recognition","authors":"Shuai Shao, Victor Sanchez","doi":"10.1109/IWBF57495.2023.10157482","DOIUrl":"https://doi.org/10.1109/IWBF57495.2023.10157482","url":null,"abstract":"Human activity recognition (HAR) is a core research topic in mobile and wearable computing, and has been applied in many applications including biometrics, health monitoring and sports coaching. In recent years, researchers have focused more attention on sensor-based HAR due to the popularity of sensor devices. However, sensor-based HAR faces the challenge of limited data size caused by the high cost of data collection and labelling work, resulting in low performance for HAR tasks. Data transformation and generative adversarial network (GAN) have been proposed as data augmentation approaches to enrich sensor data, thereby addressing the problem of data size limitations. In this paper, we studied the effectiveness of diffusion-based generative models for generating synthetic sensor data as compared to the other data augmentation approaches in sensor-based HAR. In addition, UNet has been redesigned in order to improve the efficiency and practicality of diffusion modelling. Experiments on two public datasets showed the performance of diffusion modelling compared with different data augmentation methods, indicating the feasibility of synthetic sensor data generated using diffusion modelling.","PeriodicalId":273412,"journal":{"name":"2023 11th International Workshop on Biometrics and Forensics (IWBF)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114380231","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-04-19DOI: 10.1109/IWBF57495.2023.10157540
Rod Izadi, Chen Liu
Face in Video Recognition (FiVR) commonly follows a sequential pipeline of face detection, face quality assessment, and face recognition. However, performing these often machine learning-based tasks sequentially in real-time is a challenge when considering the excessive overhead caused by convolution and other feature extraction operations typically seen in neural networks employed across these stages. To overcome this challenge, a process that can perform these operations in parallel is needed. In this paper, we propose a methodology that can alleviate the constraints of real-time processing found in the sequential pipeline of FiVR. We exploit the similarities in features used in face detection and face quality assessment, hence designing a multi-tasked face detection and quality assessment network which can perform our FiVR operations with less inference time without sparing prediction accuracy. We evaluated the face quality prediction performance of our proposed approach in comparison with a stand-alone face quality network. We also evaluated the reduction in inference time by comparing the prediction speed of our multi-tasked face detection and quality network against its sequential counterparts. Our experimental results show that our multi-tasked model can successfully meet real-time processing demand while performing at the same level of accuracy as the sequential stand-alone models.
{"title":"A Multi-Tasked Approach Towards Face Detection and Face Quality Assessment","authors":"Rod Izadi, Chen Liu","doi":"10.1109/IWBF57495.2023.10157540","DOIUrl":"https://doi.org/10.1109/IWBF57495.2023.10157540","url":null,"abstract":"Face in Video Recognition (FiVR) commonly follows a sequential pipeline of face detection, face quality assessment, and face recognition. However, performing these often machine learning-based tasks sequentially in real-time is a challenge when considering the excessive overhead caused by convolution and other feature extraction operations typically seen in neural networks employed across these stages. To overcome this challenge, a process that can perform these operations in parallel is needed. In this paper, we propose a methodology that can alleviate the constraints of real-time processing found in the sequential pipeline of FiVR. We exploit the similarities in features used in face detection and face quality assessment, hence designing a multi-tasked face detection and quality assessment network which can perform our FiVR operations with less inference time without sparing prediction accuracy. We evaluated the face quality prediction performance of our proposed approach in comparison with a stand-alone face quality network. We also evaluated the reduction in inference time by comparing the prediction speed of our multi-tasked face detection and quality network against its sequential counterparts. Our experimental results show that our multi-tasked model can successfully meet real-time processing demand while performing at the same level of accuracy as the sequential stand-alone models.","PeriodicalId":273412,"journal":{"name":"2023 11th International Workshop on Biometrics and Forensics (IWBF)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129647894","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：