Figure - available from: Multimedia Tools and Applications
This content is subject to copyright. Terms and conditions apply.
Source publication
Deep convolutional neural networks (CNNs) are widely used in face recognition, because they can extract features with higher discrimination, which is the basis for correctly identifying the identity of a face image. In order to improve the face recognition performance, in addition to improving the structures of convolutional neural networks, many n...
Similar publications
High-resolution magnetic resonance images can provide fine-grained anatomical information, but acquiring such data requires a long scanning time. In this paper, a framework called the Fused Attentive Generative Adversarial Networks(FA-GAN) is proposed to generate the super-resolution MR image from low-resolution magnetic resonance images, which can...
Remote sensing scene classification (RSSC) is a very crucial subtask of remote sensing image understanding. With the rapid development of convolutional neural networks (CNNs) in the field of natural images, great progress has been made in RSSC. Compared with natural images, labeled remote sensing images are more difficult to acquire, and typical RS...
Defect segmentation in industrial quality control is pivotal for maintaining high product surface standards, but it remains a challenge due to complex defect patterns and varying backgrounds. Traditional segmentation methods often struggle with sensitivity and specificity, leading to either missed defects or false positives. In order to address the...
![The calculation process for LX\documentclass[12pt]{minimal}...](publication/366549847/figure/fig3/AS:11431281171585577@1688263177107/The-calculation-process-for-LXdocumentclass12ptminimal-usepackageamsmath_Q320.jpg)
![The calculation process of LG(Y)\documentclass[12pt]{minimal}...](publication/366549847/figure/fig4/AS:11431281171594950@1688263177195/The-calculation-process-of-LGYdocumentclass12ptminimal-usepackageamsmath_Q320.jpg)
Recently, single image super-resolution (SISR) has made great progress due to the rapid development of deep convolutional neural networks (CNN), and the application of Generative Adversarial Networks (GAN ) has made super-resolution networks even more effective. However, GAN-based methods have many problems such as lengthy and unstable convergence....
Anomaly detection, which is approaching the problem of face anti-spoofing as a one-class classification problem, is emerging as an increasingly popular alternative to the traditional approach of training binary classifiers on specialized anti-spoofing databases which contain both client and imposter samples. In this paper, we discuss the training p...
Citations
... This method, known as Deep Metric Learning(DML), has resulted in substantial performance improvements [32], opening a new frontier in the field. By leveraging the ability to learn non-linear feature representations, deep metric learning has demonstrated outstanding results across a wide range of tasks, including highlight detection [30], [66], zero-shot image classification [18], [63], clustering [3], image retrieval [13], [19], [49], [50], [54], visual product search [8], [10], [33], [34], [55], [56], face recognition [5], [17], [25], [28], [42], [47], [64], person re-identification [4], [57], feature matching [46], fine-grained image classification [16], [45], zero-shot learning [35], [43], and collaborative filtering [36]. Moreover, deep metric learning has been successfully applied to 3D retrieval tasks [23]. ...
... To address this, several adaptive methods were introduced. For instance, AdaptiveFace [61] and Dyn-arcFace [36] focus on class imbalances by learning adaptive margins for each class, providing better handling of underrepresented classes. CurricularFace [35], on the other hand, uses curriculum learning to gradually adjust the balance between easy and hard examples during training, further refining the learning process. ...
Multi-modal deep metric learning is crucial for effectively capturing diverse representations in tasks such as face verification, fine-grained object recognition, and product search. Traditional approaches to metric learning, whether based on distance or margin metrics, primarily emphasize class separation, often overlooking the intra-class distribution essential for multi-modal feature learning. In this context, we propose a novel loss function called Density-Aware Adaptive Margin Loss(DAAL), which preserves the density distribution of embeddings while encouraging the formation of adaptive sub-clusters within each class. By employing an adaptive line strategy, DAAL not only enhances intra-class variance but also ensures robust inter-class separation, facilitating effective multi-modal representation. Comprehensive experiments on benchmark fine-grained datasets demonstrate the superior performance of DAAL, underscoring its potential in advancing retrieval applications and multi-modal deep metric learning.
... AdaptiveFace [20] and Dyn-arcFace [13] address the issue of unbalanced data by learning an adaptive margin for each class, while CurricularFace [11] adaptively adjusts the relative importance of easy and hard samples during different training stages by using curriculum learning with the loss function. On the other hand, KappaFace [25] adaptively modulates the positive margins based on class imbalance and difficulty. ...
Deep feature learning has become crucial in large-scale face recognition, and margin-based loss functions have demonstrated impressive success in this field. These methods aim to enhance the discriminative power of the softmax loss by increasing the feature margin between different classes. These methods assume class balance, where a fixed margin is sufficient to squeeze intra-class variation equally. However, real-face datasets often exhibit imbalanced classes, where the fixed margin is suboptimal, limiting the discriminative power and generalizability of the face recognition model. Furthermore, margin-based approaches typically focus on enhancing discrimination either in the angle or cosine space, emphasizing one boundary while disregarding the other. To overcome these limitations, we propose a joint adaptive margins loss function (JAMsFace) that learns class-related margins for both angular and cosine spaces. This approach allows adaptive margin penalties to adjust adaptively for different classes. We explain and analyze the proposed JAMsFace geometrically and present comprehensive experiments on multiple face recognition benchmarks. The results show that JAMsFace outperforms existing face recognition losses in mainstream face recognition tasks. Specifically, JAMsFace advances the state-of-the-art face recognition performance on LFW, CPLFW, and CFP-FP and achieves comparable results on CALFW and AgeDB-30. Furthermore, for the challenging IJB-B and IJB-C benchmarks, JAMsFace achieves impressive true acceptance rates (TARs) of 89.09% and 91.81% at a false acceptance rate (FAR) of 1e-4, respectively.
... Later, ArcFace [4] proposed additive angular margin by deploying angular penalty margin on the angle between the deep features and their corresponding weights. The great success of softmax loss with penalty margin motivated several works to propose a novel variant of softmax loss [11], [14], [5], [13], [27], [10], [19], [1]. All these solutions achieved notable accuracies on mainstream benchmarks [9], [25], [29], [18] for face recognition. ...
... The proposed loss targets the easy samples at an early stage of training and the hard ones at a later stage of training. Jiao et al. [11] proposed Dyn-arcface based on ArcFace loss [4] by replacing the fixed margin value of ArcFace with an adaptive one. The margin value of Dyn-arcface is adjusted based on the distance between each class center and the other class centers. ...
Learning discriminative face features plays a major role in building high-performing face recognition models. The recent state-of-the-art face recognition solutions proposed to incorporate a fixed penalty margin on commonly used classification loss function, softmax loss, in the normalized hypersphere to increase the discriminative power of face recognition models, by minimizing the intra-class variation and maximizing the inter-class variation. Marginal softmax losses, such as ArcFace and CosFace, assume that the geodesic distance between and within the different identities can be equally learned using a fixed margin. However, such a learning objective is not realistic for real data with inconsistent inter-and intra-class variation, which might limit the discriminative and generalizability of the face recognition model. In this paper, we relax the fixed margin constrain by proposing elastic margin loss (ElasticFace) that allows flexibility in the push for class separability. The main idea is to utilize random margin values drawn from a normal distribution in each training iteration. This aims at giving the margin chances to extract and retract to allow space for flexible class separability learning. We demonstrate the superiority of our elastic margin loss over ArcFace and CosFace losses, using the same geometric transformation, on a large set of mainstream benchmarks. From a wider perspective, our ElasticFace has advanced the state-of-the-art face recognition performance on six out of nine mainstream benchmarks.
Face attribute recognition is also widely used in human–computer interaction, train stations, and other fields because face attributes are rich in information. A convolutional neural network and distributed computing (DC)-based face recognition model is researched to enhance multitask face recognition accuracy and computational capacity. The model is also trained using a multitask learning method for face identity recognition, fatigue state recognition, age recognition, and gender recognition. The model is streamlined by DC to improve the computational efficiency of the model. To further raise the recognition accuracy of each task, the study combines the add feature fusion (FF) principle and concat FF principle to propose an improved face recognition model. The outcomes of the study revealed that the true acceptance rate value of the model reached 0.95, and the face identity recognition accuracy reached 100%. The fatigue state determination accuracy reached 99.12%. The average recognition time of a single photo was 354 ms. The model can quickly recognize face identity and face attribute information in a short time, with short time and high recognition accuracy. It is beneficial in several areas, including intelligent navigation and driving behavior analysis.
This study aims to optimize the interior design generation process while enhancing personalization and efficiency through the development of a hybrid Residual Network-Human-Computer Interaction (ResNet-HCI) framework. It employs Residual Network (ResNet) neural networks, which leverage residual learning to improve training stability and feature extraction capabilities in deep networks. This allows for efficient feature reuse and optimization of model performance. Additionally, Human-Computer Interaction (HCI) technologies, such as voice commands, gesture control, and Virtual Reality/Augmented Reality (VR/AR), are integrated to enhance user interaction with the design system, thereby improving the intelligence and personalization of the interior design workflow. The Large-Scale Scene Understanding dataset is used for model training to evaluate system performance under varying training steps, hyperparameter configurations, and noise conditions. The experimental results show the following: (1) Significant performance variations are observed across different models under conditions such as increased training iterations, noise interference, and design scoring. In the iteration experiment, model performance generally improves with more training steps. ResNet50 consistently outperforms other models, achieving an F1 score of 0.935 after 20 iterations, demonstrating exceptional feature learning and stability. In the noise robustness analysis, ResNet50 and ResNeXt show minimal performance degradation under Gaussian noise, indicating strong noise robustness. (2) Regarding interior design scoring, hybrid layouts generated using ResNet models combined with HCI technologies excel in multiple dimensions, achieving the highest overall satisfaction scores and emerging as the optimal design solutions. These findings validate the exceptional performance and versatility of ResNet50 and ResNeXt in both deep learning and HCI applications. This study provides both theoretical and practical support for the intelligent transformation of the interior design field, while also offering insights into broader applications in creative industries.
Home textile images have small inter-class differences and large intra-class differences, making home textile image retrieval face great technical challenges. In this paper, we design an effective retrieval model for home textile images, in which ResNet50 is used as backbone network, and a hybrid maximal pooling spatial attention module is proposed to fuse local spatial information at different scales, thus focusing on key information and suppressing irrelevant information. Moreover, we propose a new loss function called SD-arcface for fine-grained feature recognition, which adopts dynamic additive angular margin to improve the intra-class compactness and the inter-class separation of home textile images. In addition, we set up a large-scale dataset of home textile images, which contains 89k home textile images from 12k categories, and evaluate the image retrieval performance of the proposed model with two metrics, Recall@k and MAP@k. Finally, the experimental results show that the proposed model achieves a better retrieval performance than other models.
Deep learning‐based face recognition models have demonstrated remarkable performance in benchmark tests, and knowledge distillation technology has been frequently accustomed to obtain high‐precision real‐time face recognition models specifically designed for mobile and embedded devices. However, in recent years, the knowledge distillation methods for face recognition, which mainly focus on feature or logit knowledge distillation techniques, neglect the attention mechanism that play an important role in the domain of neural networks. An innovation cross‐stage connection review path of the attention cosine similarity knowledge distillation method that unites the attention mechanism with review knowledge distillation method is proposed. This method transfers the attention map obtained from the teacher network to the student through a cross‐stage connection path. The efficacy and excellence of the proposed algorithm are demonstrated in popular benchmark tests.
Feature extractors significantly impact the performance of biometric systems. In the field of hand gesture authentication, existing studies focus on improving the model architectures and behavioral characteristic representation methods to enhance their feature extractors. However, loss functions, which can guide extractors to produce more discriminative identity features, are neglected. In this paper, we improve the margin-based Softmax loss functions, which are mainly designed for face authentication, in two aspects to form a new loss function for hand gesture authentication. First, we propose to replace the commonly used cosine function in the margin-based Softmax losses with a linear function to measure the similarity between identity features and proxies (the weight matrix of Softmax, which can be viewed as class centers). With the linear function, the main gradient magnitude decreases monotonically as the quality of the model improves during training, thus allowing the model to be quickly optimized in the early stage and precisely fine-tuned in the late stage. Second, we design an adaptive margin scheme to assign margin penalties to different samples according to their separability and the model quality in each iteration. Our adaptive margin scheme constrains the gradient magnitude. It can reduce radical (excessively large) gradient magnitudes and provide moderate (not too small) gradient magnitudes for model optimization, contributing to more stable training. The linear function and the adaptive margin scheme are complementary. Combining them, we obtain the proposed linear adaptive additive angular margin (L3AM) loss. To demonstrate the effectiveness of L3AM loss, we conduct extensive experiments on seven hand-related authentication datasets, compare it with 25 state-of-the-art (SOTA) loss functions, and apply it to eight SOTA hand gesture authentication models. The experimental results show that L3AM loss further improves the performance of the eight authentication models and outperforms the 25 losses. The code is available at https://github.com/SCUT-BIP-Lab/L3AM.
