Cross-dataset Scenarios Research Articles

Towards Robust Person Re-Identification by Defending Against Universal Attackers.

Recent studies show that deep person re-identification (re-ID) models are vulnerable to adversarial examples, so it is critical to improving the robustness of re-ID models against attacks. To achieve this goal, we explore the strengths and weaknesses of existing re-ID models, i.e., designing learning-based attacks and training robust models by defending against the learned attacks. The contributions of this paper are three-fold: First, we build a holistic attack-defense framework to study the relationship between the attack and defense for person re-ID. Second, we introduce a combinatorial adversarial attack that is adaptive to unseen domains and unseen model types. It consists of distortions in pixel and color space (i.e., mimicking camera shifts). Third, we propose a novel virtual-guided meta-learning algorithm for our attack-defense system. We leverage a virtual dataset to conduct experiments under our meta-learning framework, which can explore the cross-domain constraints for enhancing the generalization of the attack and the robustness of the re-ID model. Comprehensive experiments on three large-scale re-ID benchmarks demonstrate that: 1) Our combinatorial attack is effective and highly universal in cross-model and cross-dataset scenarios; 2) Our meta-learning algorithm can be readily applied to different attack and defense approaches, which can reach consistent improvement; 3) The defense model trained on the learning-to-learn framework is robust to recent SOTA attacks that are not even used during training.

Improving Transfer Performance of Deep Learning with Adaptive Batch Normalization for Brain-computer Interfaces.

Recently, transfer learning and deep learning have been introduced to solve intra- and inter-subject variability problems in Brain-Computer Interfaces. However, the generalization ability of these BCIs is still to be further verified in a cross-dataset scenario. This study compared the transfer performance of manifold embedded knowledge transfer and pre-trained EEGNet with three preprocessing strategies. This study also introduced AdaBN for target domain adaptation. The results showed that EEGNet with Riemannian alignment and AdaBN could achieve the best transfer accuracy about 65.6% on the target dataset. This study may provide new insights into the design of transfer neural networks for BCIs by separating source and target batch normalization layers in the domain adaptation process.

Impact of Lung Segmentation on the Diagnosis and Explanation of COVID-19 in Chest X-ray Images.

COVID-19 frequently provokes pneumonia, which can be diagnosed using imaging exams. Chest X-ray (CXR) is often useful because it is cheap, fast, widespread, and uses less radiation. Here, we demonstrate the impact of lung segmentation in COVID-19 identification using CXR images and evaluate which contents of the image influenced the most. Semantic segmentation was performed using a U-Net CNN architecture, and the classification using three CNN architectures (VGG, ResNet, and Inception). Explainable Artificial Intelligence techniques were employed to estimate the impact of segmentation. A three-classes database was composed: lung opacity (pneumonia), COVID-19, and normal. We assessed the impact of creating a CXR image database from different sources, and the COVID-19 generalization from one source to another. The segmentation achieved a Jaccard distance of 0.034 and a Dice coefficient of 0.982. The classification using segmented images achieved an F1-Score of 0.88 for the multi-class setup, and 0.83 for COVID-19 identification. In the cross-dataset scenario, we obtained an F1-Score of 0.74 and an area under the ROC curve of 0.9 for COVID-19 identification using segmented images. Experiments support the conclusion that even after segmentation, there is a strong bias introduced by underlying factors from different sources.

Joint optic disc and cup segmentation based on multi-scale feature analysis and attention pyramid architecture for glaucoma screening

Automatic segmentation of optic disc (OD) and optic cup (OC) is an essential task for analysing colour fundus images. In clinical practice, accurate OD and OC segmentation assist ophthalmologists in diagnosing glaucoma. In this paper, we propose a unified convolutional neural network, named ResFPN-Net, which learns the boundary feature and the inner relation between OD and OC for automatic segmentation. The proposed ResFPN-Net is mainly composed of multi-scale feature extractor, multi-scale segmentation transition and attention pyramid architecture. The multi-scale feature extractor achieved the feature encoding of fundus images and captured the boundary representations. The multi-scale segmentation transition is employed to retain the features of different scales. Moreover, an attention pyramid architecture is proposed to learn rich representations and the mutual connection in the OD and OC. To verify the effectiveness of the proposed method, we conducted extensive experiments on two public datasets. On the Drishti-GS database, we achieved a Dice coefficient of 97.59%, 89.87%, the accuracy of 99.21%, 98.77%, and the Averaged Hausdorff distance of 0.099, 0.882 on the OD and OC segmentation, respectively. We achieved a Dice coefficient of 96.41%, 83.91%, the accuracy of 99.30%, 99.24%, and the Averaged Hausdorff distance of 0.166, 1.210 on the RIM-ONE database for OD and OC segmentation, respectively. Comprehensive results show that the proposed method outperforms other competitive OD and OC segmentation methods and appears more adaptable in cross-dataset scenarios. The introduced multi-scale loss function achieved significantly lower training loss and higher accuracy compared with other loss functions. Furthermore, the proposed method is further validated in OC to OD ratio calculation task and achieved the best MAE of 0.0499 and 0.0630 on the Drishti-GS and RIM-ONE datasets, respectively. Finally, we evaluated the effectiveness of the glaucoma screening on Drishti-GS and RIM-ONE datasets, achieving the AUC of 0.8947 and 0.7964. These results proved that the proposed ResFPN-Net is effective in analysing fundus images for glaucoma screening and can be applied in other relative biomedical image segmentation applications.

Enhancing transfer performance across datasets for brain-computer interfaces using a combination of alignment strategies and adaptive batch normalization

Objective. Recently, transfer learning (TL) and deep learning (DL) have been introduced to solve intra- and inter-subject variability problems in brain-computer interfaces (BCIs). However, current TL and DL algorithms are usually validated within a single dataset, assuming that data of the test subjects are acquired under the same condition as that of training (source) subjects. This assumption is generally violated in practice because of different acquisition systems and experimental settings across studies and datasets. Thus, the generalization ability of these algorithms needs further validations in a cross-dataset scenario, which is closer to the actual situation. This study compared the transfer performance of pre-trained deep-learning models with different preprocessing strategies in a cross-dataset scenario. Approach. This study used four publicly available motor imagery datasets, each was successively selected as a source dataset, and the others were used as target datasets. EEGNet and ShallowConvNet with four preprocessing strategies, namely channel normalization, trial normalization, Euclidean alignment, and Riemannian alignment, were trained with the source dataset. The transfer performance of pre-trained models was validated on the target datasets. This study also used adaptive batch normalization (AdaBN) for reducing interval covariate shift across datasets. This study compared the transfer performance of using the four preprocessing strategies and that of a baseline approach based on manifold embedded knowledge transfer (MEKT). This study also explored the possibility and performance of fusing MEKT and EEGNet. Main results. The results show that DL models with alignment strategies had significantly better transfer performance than the other two preprocessing strategies. As an unsupervised domain adaptation method, AdaBN could also significantly improve the transfer performance of DL models. The transfer performance of DL models that combined AdaBN and alignment strategies significantly outperformed MEKT. Moreover, the generalizability of EEGNet models that combined AdaBN and alignment strategies could be further improved via the domain adaptation step in MEKT, achieving the best generalization ability among multiple datasets (BNCI2014001: 0.788, PhysionetMI: 0.679, Weibo2014: 0.753, Cho2017: 0.650). Significance. The combination of alignment strategies and AdaBN could easily improve the generalizability of DL models without fine-tuning. This study may provide new insights into the design of transfer neural networks for BCIs by separating source and target batch normalization layers in the domain adaptation process.

On the generalisation capabilities of Fisher vector‐based face presentation attack detection

In the last decades, the broad development experienced by biometric systems has unveiled several threats which may decrease their trustworthiness. Those are attack presentations which can be easily carried out by a non-authorised subject to gain access to the biometric system. In order to mitigate those security concerns, most face Presentation Attack Detection techniques have reported a good detection performance when they are evaluated on known Presentation Attack Instruments (PAI) and acquisition conditions, in contrast to more challenging scenarios where unknown attacks are included in the test set. For those more realistic scenarios, the existing algorithms face difficulties to detect unknown PAI species in many cases. In this work, we use a new feature space based on Fisher Vectors, computed from compact Binarised Statistical Image Features histograms, which allow discovering semantic feature subsets from known samples in order to enhance the detection of unknown attacks. This new representation, evaluated for challenging unknown attacks taken from freely available facial databases, shows promising results: a BPCER100 under 17% together with an AUC over 98% can be achieved in the presence of unknown attacks. In addition, by training a limited number of parameters, our method is able to achieve state-of-the-art deep learning-based approaches for cross-dataset scenarios.

An improved faster-RCNN model for handwritten character recognition

Existing techniques for hand-written digit recognition (HDR) rely heavily on the hand-coded key points and requires prior knowledge. Training an efficient HDR network with these preconditions is a complicated task. Recently, work on HDR is mainly focused on deep learning (DL) approaches and has exhibited remarkable results. However, effective detection and classification of numerals is still a challenging task due to people’s varying writing styles and the presence of blurring, distortion, light and size variations in the input sample. To cope with these limitations, we present an effective and efficient HDR system, introducing a customized faster regional convolutional neural network (Faster-RCNN). This approach comprises three main steps. Initially, we develop annotations to obtain the region of interest. Then, an improved Faster-RCNN is employed in which DenseNet-41 is introduced to compute the deep features. Finally, the regressor and classification layer is used to localize and classify the digits into ten classes. The performance of the proposed method is analyzed on the standard MNIST database, which is diverse in terms of changes in lighting conditions, chrominance, shape and size of digits, and the occurrence of blurring and noise effects, etc. Additionally, we have also evaluated our technique over a cross-dataset scenario to prove its efficacy. Experimental evaluations demonstrate that the approach is more competent and able to accurately detect and classify numerals than other recent methods.

Cross-Dataset Point Cloud Recognition Using Deep-Shallow Domain Adaptation Network

In this work, we propose a new two-view domain adaptation network named Deep-Shallow Domain Adaptation Network (DSDAN) for 3D point cloud recognition. Different from the traditional 2D image recognition task, the valuable texture information is often absent in point cloud data, making point cloud recognition a challenging task, especially in the cross-dataset scenario where the training and testing data exhibit a considerable distribution mismatch. In our DSDAN method, we tackle the challenging cross-dataset 3D point cloud recognition task from two aspects. On one hand, we propose a two-view learning framework, such that we can effectively leverage multiple feature representations to improve the recognition performance. To this end, we propose a simple and efficient Bag-of-Points feature method, as a complementary view to the deep representation. Moreover, we also propose a cross view consistency loss to boost the two-view learning framework. On the other hand, we further propose a two-level adaptation strategy to effectively address the domain distribution mismatch issue. Specifically, we apply a feature-level distribution alignment module for each view, and also propose an instance-level adaptation approach to select highly confident pseudo-labeled target samples for adapting the model to the target domain, based on which a co-training scheme is used to integrate the learning and adaptation process on the two views. Extensive experiments on the benchmark dataset show that our newly proposed DSDAN method outperforms the existing state-of-the-art methods for the cross-dataset point cloud recognition task.

Ensemble of Multi-View Learning Classifiers for Cross-Domain Iris Presentation Attack Detection

The adoption of large-scale iris recognition systems around the world has brought to light the importance of detecting presentation attack images (textured contact lenses and printouts). This work presents a new approach in iris Presentation Attack Detection (PAD), by exploring combinations of Convolutional Neural Networks (CNNs) and transformed input spaces through binarized statistical image features (BSIF). Our method combines lightweight CNNs to classify multiple BSIF views of the input image. Following explorations on complementary input spaces leading to more discriminative features to detect presentation attacks, we also propose an algorithm to select the best (and most discriminative) predictors for the task at hand.An ensemble of predictors makes use of their expected individual performances to aggregate their results into a final prediction. Results show that this technique improves on the current state of the art in iris PAD, outperforming the winner of LivDet-Iris2017 competition both for intra- and cross-dataset scenarios, and illustrating the very difficult nature of the cross-dataset scenario.

Joint Discriminative Learning of Deep Dynamic Textures for 3D Mask Face Anti-Spoofing

Three-dimensional mask spoofing attacks have been one of the main challenges in face recognition. Compared with a 3D mask, a real face displays different facial motion patterns that are reflected by different facial dynamic textures. However, a large portion of these facial motion differences is subtle. We find that the subtle facial motion can be fully captured by multiple deep dynamic textures from a convolutional layer of a convolutional neural network, but not all deep dynamic textures from different spatial regions and different channels of a convolutional layer are useful for differentiation of subtle motions between real faces and 3D masks. In this paper, we propose a novel feature learning model to learn discriminative deep dynamic textures for 3D mask face anti-spoofing. A novel joint discriminative learning strategy is further incorporated in the learning model to jointly learn the spatial- and channel-discriminability of the deep dynamic textures. The proposed joint discriminative learning strategy can be used to adaptively weight the discriminability of the learned feature from different spatial regions or channels, which ensures that more discriminative deep dynamic textures play more important roles in face/mask classification. Experiments on several publicly available data sets validate that the proposed method achieves promising results in intra- and cross-data set scenarios.

On the design of automatic voice condition analysis systems. Part II: Review of speaker recognition techniques and study on the effects of different variability factors

Abstract This is the second of a two-part series devoted to the automatic voice condition analysis of voice pathologies, being a direct continuation to the paper On the design of automatic voice condition analysis systems. Part I: review of concepts and an insight to the state of the art. The aim of this study is to examine several variability factors affecting the robustness of systems that automatically detect the presence of voice pathologies by means of audio registers. Multiple experiments are performed to test out the influence of the speech task, extralinguistic aspects (such as sex), the acoustic features and the classifiers in their performance. Some experiments are carried out using state-of-the-art classification methodologies often employed in speaker recognition. In order to evaluate the robustness of the methods, testing is repeated across several corpora with the aim to create a single system integrating the conclusions obtained previously. This system is later tested under cross-dataset scenarios in an attempt to obtain more realistic conclusions. Results identify a reduced subset of relevant features, which are used in a hierarchical-like scenario incorporating information of different speech tasks. In particular, for the experiments carried out using the Saarbruecken voice dataset, the area under the ROC curve of the system reached 0.88 in an intra-dataset setting and ranged from 0.82 to 0.94 in cross-dataset scenarios. These results let us open a discussion about the suitability of these techniques to be transferred to the clinical setting.

Person Reidentification Using Deep Convnets With Multitask Learning

Person reidentification involves recognizing a person across nonoverlapping camera views, with different pose, illumination, and camera characteristics. We propose to tackle this problem by training a deep convolutional network to represent a person’s appearance as a low-dimensional feature vector that is invariant to common appearance variations encountered in the reidentification problem. Specifically, a Siamese network architecture is used to train a feature extraction network using pairs of similar and dissimilar images. We show that the use of a novel multitask learning objective is crucial for regularizing the network parameters in order to prevent overfitting due to the small size of the training data set. We complement the verification task, which is at the heart of reidentification, by training the network to jointly perform verification and identification and to recognize attributes related to the clothing and pose of the person in each image. In addition, we show that our proposed approach performs well even in the challenging cross-data set scenario, which may better reflect real-world expected performance.

Joint Patch and Multi-label Learning for Facial Action Unit and Holistic Expression Recognition.

Most action unit (AU) detection methods use one-versus-all classifiers without considering dependences between features or AUs. In this paper, we introduce a joint patch and multi-label learning (JPML) framework that models the structured joint dependence behind features, AUs, and their interplay. In particular, JPML leverages group sparsity to identify important facial patches, and learns a multi-label classifier constrained by the likelihood of co-occurring AUs. To describe such likelihood, we derive two AU relations, positive correlation and negative competition, by statistically analyzing more than 350,000 video frames annotated with multiple AUs. To the best of our knowledge, this is the first work that jointly addresses patch learning and multi-label learning for AU detection. In addition, we show that JPML can be extended to recognize holistic expressions by learning common and specific patches, which afford a more compact representation than the standard expression recognition methods. We evaluate JPML on three benchmark datasets CK+, BP4D, and GFT, using within-and cross-dataset scenarios. In four of five experiments, JPML achieved the highest averaged F1 scores in comparison with baseline and alternative methods that use either patch learning or multi-label learning alone.

Cross Dataset Evaluation of Feature Extraction Techniques for Leaf Classification

In this work feature extraction techniques for leaf classification are evaluated in a cross dataset scenario. First, a leaf identification system consisting of six feature classes is described and tested on five established publicly available datasets by using standard evaluation procedures within the datasets. Afterwards, the performance of the developed system is evaluated in the much more challenging scenario of cross dataset evaluation. Finally, a new dataset is introduced as well as a web service, which allows to identify leaves both photographed on paper and when still attached to the tree. While the results obtained during classification within a dataset come close to the state of the art, the classification accuracy in cross dataset evaluation is significantly worse. However, by adjusting the system and taking the top five predictions into consideration very good results of up to 98% are achieved. It is shown that this difference is down to the ineffectiveness of certain feature classes as well as the increased severity of the task as leaves that grew under different environmental influences can differ significantly not only in colour, but also in shape.