The increasing need for robust digital signature verification systems has amplified interest in realistic online signature generation to counter digital forgeries. In this work, we propose a novel Decision Transformer based framework that learns Reinforcement Learning output to generate diverse online signatures. Departing from traditional RL approaches that rely on policy gradients or value function estimation, we formulate signature generation as a sequence-modeling problem. Our framework addresses varied free-form signature styles, demonstrating adaptability across linguistic and stylistic variations. Initially, an RL model generates signature trajectories, which are then fed to Decision Transformer, employing an autoregressive sequence modeling approach. To further personalize the generated signatures, we introduce a Q-learning-based module that produces user-specific variations while mitigating noise. By operating in an offline reinforcement learning setting, the proposed method reduces the dependency on extensive online interactions, improving scalability. Experimental results on a publicly available online signature dataset in multiple linguistic script styles show that our approach significantly outperforms traditional generative methods in terms of realism, variability, and mimicry accuracy. These results highlight the potential of Decision Transformers for structured sequence generation tasks beyond their conventional domains. Decision Transforme

Scoliosis presents significant diagnostic challenges, especially in its early stages due to structural deformities of the spine. The most common type, Adolescent Idiopathic Scoliosis (AIS), typically appears during rapid growth periods between ages 10 and 15, accounting for about 80–85% of all scoliosis cases. Currently, diagnosis mainly involves repeated X-rays and clinical evaluations, which are costly and expose patients to frequent radiation. To address these challenges, we propose ScoliGaitX, a novel non-invasive system for assessing scoliosis through gait video analysis. Our approach uses a multi-modal deep learning model trained specifically for scoliosis classification using the Scoliosis1K dataset. We integrate three different gait modalities—(i) silhouettes to capture appearance, (ii) optical flow for motion analysis, and (iii) GEI-subtract sequences to highlight deviations from normal gait—to effectively identify gait abnormalities associated with scoliosis. A key component of our proposal is the Align Gate Fusion (AGF) module, designed to efficiently learn relationships between different modalities. It achieves this by adaptively assigning importance to each modality through a lightweight global weighted-fusion mechanism. Our experimental results demonstrate that ScoliGaitX significantly outperforms existing methods, achieving an accuracy of 89.05%, which is 7.05% higher than the previous best approach (ScolNet-MT), while maintaining excellent specificity. This highlights the promise of our method for providing early, radiation-free scoliosis assessment.

Identity details within an ECG is jointly situated within local and global features. The current methods for ECG recognition emphasize only on local or global details. They have also paid limited attention to unseen and cross-domain scenarios. Furthermore, there exists a lack of consensus on evaluation strategies. Thus, this paper introduces LGTraNet, a generalized architecture designed to establish baselines for securing personal identity using ECG biometrics in cross-domain scenarios. Our proposed model firstly extracts identity details at local temporal levels. The extracted features are then calibrated with globally details using a Self-Calibrated Normalizing Residual Network (SCNRNet). Finally, the refined local details are aggregated using a transformer model to formulate robust global identity representations. We evaluate LGTraNet over challenging cross-domain scenarios, such as cross-session and cross-database. To mitigate challenges in domain-shift, we also introduce an transfer learning based training strategy. Experimental study conducted on three benchmark datasets, ECG1D, MIT-BIH, and PTB, shows that the LGTraNet achieves significant performance in cross-domain settings, and outperforms state-of-the-art. Our code is available at: https://github.com/AmanVerma2307/LGTraNet.

LivDet-Iris 2025 is the sixth edition of the iris liveness detection competition. Held every two to three years, the competition aims to foster the development of robust algorithms capable of detecting a wide range of physically-and digitally-presented attacks in iris biometrics. The 2025 edition obtained the largest number of submissions in the history of the competition: ten algorithms from five institutions, and one commercial iris recognition system. LivDet-Iris 2025 also introduced new tasks compared to previous editions: (Task 1) a benchmark offered by an industry partner, (Task 2) morphed iris images, in which two different-identity samples were blended into one image, and (Task 3) evaluation of presentation attack detection robustness against advanced manufacturing techniques for textured contact lenses. This edition, for the first time in the series, offers a systematic testing of a commercial iris recognition system (software and hardware) using physical artifacts presented to the sensor. Dermalog-Iris team submitted algorithms that won all tasks, achieving the area under the ROC curve of 90.57%, 68.23% and 99.99% in tasks 1, 2, and 3, respectively. Additionally, we include results for baseline algorithms, based on modern deep convolutional neural networks and trained with all available public datasets of iris images representing bona fide samples and anomalies (physical attacks, eye diseases, post-mortem cases, and synthetically-generated iris images). Test samples created for tasks 2 and 3, and baseline models are made available to offer the state-of-the-art benchmark for iris liveness detection.

Advancements in virtual and head-mounted devices have introduced new challenges for iris biometrics, such as varying gaze directions, partial occlusions, and inconsistent lighting conditions. To address these obstacles, we present VREyeSAM, a robust iris segmentation framework specifically designed for images captured under both steady and dynamic gaze scenarios. Our pipeline includes a quality-aware pre-processing module that filters out partially or fully closed eyes, ensuring that only high-quality, fully open iris images are used for training and inference. In addition, we introduce an uncertainty weighted hybrid loss function that adaptively balances multiple learning objectives, enhancing the robustness of the model under diverse visual conditions. Using this approach, we evaluate VREyeSAM on the VRBiom dataset, where it achieves state-of-the-art performance with a Precision of 0.751, Recall of 0.870, F1-Score of 0.806, and a mean IoU of 0.647, significantly outperforming existing segmentation methods.

This paper presents a summary of the 2025 Sclera Segmentation Benchmarking Competition (SSBC), which focused on the development of privacy-preserving sclera-segmentation models trained using synthetically generated ocular images. The goal of the competition was to evaluate how well models trained on synthetic data perform in comparison to those trained on real-world datasets. The competition featured two tracks: (i) one relying solely on synthetic data for model development, and (ii) one combining/mixing synthetic with (a limited amount of) real-world data. A total of nine research groups submitted diverse segmentation models, employing a variety of architectural designs, including transformer-based solutions, lightweight models, and segmentation networks guided by generative frameworks. Experiments were conducted across three evaluation datasets containing both synthetic and real-world images, collected under diverse conditions. Results show that models trained entirely on synthetic data can achieve competitive performance, particularly when dedicated training strategies are employed, as evidenced by the top performing models that achieved F1scores of over 0.8 in the synthetic data track. Moreover, performance gains in the mixed track were often driven more by methodological choices rather than by the inclusion of real data, highlighting the promise of synthetic data for privacy-aware biometric development. The code and data for the competition is available at: https://github.com/dariant/SSBC_2025.

Early detection of scoliosis in youth enables to start treatments while the spine curve is still minor. Video-based gait analysis offers automated diagnosis and follow-ups, addressing limitations of x-ray scans and clinical assessments. Inspired by the recent success of a video-based scoliosis classification method, we propose an improved method by fusing the fine and global dynamic information of person-specific gait pose variation in a video. We postulate this to be the key to an efficient representation to discriminate a scoliosis-affected gait from a healthy one. Our proposed dynamism reinforced gait encoder ( DRGait) is a layered architecture of video feature embedding blocks comprising a motion embedding module (MEM) and a dynamism reinforcement module (DRM). MEM is a CNN encoder that extracts granular motion from frame-wise pose variation and intense motion or dynamism ingrained in the summarized pose variation. DRM is a linear operation-based module that fuses these multi-level motions to create discriminating gait embeddings. The model classifies a sample video through a prior level of individual identification. DRGait is trained on four train sets - 112, 114, 118, and 1116 in Scoliosis1K dataset [1], where respective sets contain 2, 4, 8, and 16 times higher number of scoliosis-negative subjects than Scoliosis-affected and intermediate-stage subjects. Our model achieves 88.7% and 78.3% accuracies, exceeding the work [1] by large margins on highly imbalanced sample distributions, 118 and 1116, respectively. We also achieve comparative results on moderately imbalanced sample distributions, 112 and 114. The experimental results of DRGait show the strength of reinforcing the subtle motion patterns by global motion features. This approach ensures consistent performance, regardless of class imbalance, and robustness against input perturbations. Our trained model and code will be released later.

Low-dose computed tomography (LDCT) scans have become a widely adopted technique to minimize radiation exposure in medical imaging. However, a lower radiation dose introduces noise, which degrades image quality and compromises diagnostic precision. The denoising of LDCT scans remains a critical challenge in the field of computed tomography (CT) imaging research. To address this, efficient denoising techniques such as deep learning methods have been proposed for LDCT denoising, primarily using high-dose CT (HDCT) images as ground truth. However, these methods often lead to over-smoothing and fail to capture the inherent spatial correlations within a single CT slice, particularly the anatomical semantics. To address these challenges, we propose NoisConFuse-Net, a robust deep learning framework for LDCT denoising. The core architecture features a Perturbed Encoder with Complementary Dual Decoders (PE-CDD), designed around efficient self-attention (ESA) mechanism to capture the local as well as global dependencies in the slice. Additionally, a Noise-Augmented Feature Maps (NAFM) module is introduced to enhance network generalization for image denoising by perturbing intermediate feature representations within the encoder. Our framework employs a dual decoder network for complementary learning strategy that effectively disentangles and balances noise suppression and content preservation. Further, we integrate a pretrained CLIP ResNet image encoder as a contrastive regularizer (CR) to refine the fused output from the PE-CDD, ensuring alignment between the denoised images and HDCT ground truth. This contrastive regularization preserves key anatomical structures while effectively differentiating noise in LDCT images. This novel approach outperforms the state-of-the-art methods, demonstrating its effectiveness for high-quality LDCT denoising.

The growing prevalence of AI-generated forgeries has intensified challenges in verifying handwritten signatures, a widely accepted form of biometric authentication. Existing state-of-the-art technologies have matured to authenticate signatures with high fidelity, there is now a pressing need to develop systems that can robustly detect and identify deepfake or AI-generated signatures. These synthetic forgeries, if undetected, can severely compromise security in digital systems. Therefore, to build more resilient verification systems, it is critical to create system which can learn discriminative features that uniquely characterize deepfake signatures. This paper presents Penformer, an online signature authentication system that detects both manual and AI generated forgeries. Combining structural features with log normal modeling of stroke and behavior dynamics with a multimodal transformer architecture, Penformer captures key spatial and temporal features. Experiments on openly available benchmark datasets and synthetically created datasets demonstrate SOTA results and show systems’ adaptability and effectiveness of digital signature verification against advanced forgery methods.

Recent studies have emphasized the potential of forehead-crease patterns as an alternative for face, iris, and periocular recognition, presenting contactless and convenient solutions, particularly in situations where faces are covered by surgical masks. However, collecting forehead data presents challenges, including cost and time constraints, as developing and optimizing forehead verification methods requires a substantial number of high-quality images. To tackle these challenges, the generation of synthetic biometric data has gained traction due to its ability to protect privacy while enabling effective training of deep learning-based biometric verification methods. In this paper, we present a new framework to synthesize forehead-crease image data while maintaining important features, such as uniqueness and realism. The proposed framework consists of two main modules: a Subject-Specific Generation Module (SSGM), based on an image-to-image Brownian Bridge Diffusion Model (BBDM), which learns a one-to-many mapping between image pairs to generate identity-aware synthetic forehead creases corresponding to real subjects, and a Subject-Agnostic Generation Module (SAGM), which samples new synthetic identities with assistance from the SSGM. We evaluate the diversity and realism of the generated forehead-crease images primarily using the Fréchet Inception Distance (FID) and the Structural Similarity Index Measure (SSIM). In addition, we assess the utility of synthetically generated forehead-crease images using a forehead-crease verification system (FHCVS). The results indicate an improvement in the verification accuracy of the FHCVS by utilizing synthetic data.

Signature being the most frequently used and widely accepted biometric has always attracted researchers due to complexities attached to its verification and analysis tasks. It can be categorized as offline or online based on the acquisition process. Online techniques are capable of capturing behavioral information. Signature possesses a unique spatial as well as temporal relation but suffers from large intra-class variation. Online signature verification techniques have been shown to yield better performance with respect to both intra-class variation as well as inter-class variation (forgery). This paper presents an online signature verification method where we propose: (i) a novel representation of the online signature data to video frames while maintaining structural and temporal information. (ii) a novel framework of spatio-temporal analysis for generated video frames to detect signature forgery. In this approach a method to extract long range temporal dependency is introduced which gives enhanced temporal context for better signature verification. We have conducted a series of experiments to train and validate our approach on the publicly available online signature datasets. Our proposed method outperforms the state-of-the-art, in online signature verification techniques and achieves an Equal Error Rate (EER) of 1.65% for skilled forgeries and 0.76% for random forgeries.

We investigate the reasons for lower performance of transformer based pedestrian detection models compared to convolutional neural network (CNN) based ones. CNN models generate dense pedestrian proposals, refine each proposal individually, and follow it up with non-maximal-suppression (NMS) to generate sparse predictions. In contrast, transformer models select one proposal per groundtruth (GT) pedestrian box and backpropagate positive gradient from them. All other proposals, many of them highly similar to the selected ones, are passed negative gradient. Though this leads to sparse predictions, obviating the need of NMS, the arbitrary selection of one among many similar proposals, hinders effective training, and lower accuracy of pedestrian detection. To mitigate the problem, instead of commonly used Kuhn-Munkres matching algorithm, we propose Min-cost-flow based formulation, and incorporate constraints such as, each ground truth box is matched to atleast one proposal, and many equally good proposals can be matched to a single ground truth box. We propose first transformer based pedestrian detection model incorporating our matching algorithm. Extensive experiments reveal that our approach achieves a miss rate (lower is better) of 3.7 / 17.4 / 21.8 / 8.3 / 2.0 on Eurocity / TJU-traffic / TJUcampus / Cityperson / Caltech datasets compared to 4.7 /18.7 / 24.8 / 8.5 / 3.1 by the current SOTA. Code is available at https://ajayshastry08.github.io/flow_matcher

This paper presents the summary of the Sclera Segmentation and Joint Recognition Benchmarking Competition (SSRBC 2023) held in conjunction with IEEE International Joint Conference on Biometrics (IJCB 2023). Different from the previous editions of the competition, SSRBC 2023 not only explored the performance of the latest and most advanced sclera segmentation models, but also studied the impact of segmentation quality on recognition performance. Five groups took part in SSRBC 2023 and submitted a total of six segmentation models and one recognition technique for scoring. The submitted solutions included a wide variety of conceptually diverse deep-learning models and were rigorously tested on three publicly available datasets, i.e., MASD, SBVPI and MOBIUS. Most of the segmentation models achieved encouraging segmentation and recognition performance. Most importantly, we observed that better segmentation results always translate into better verification performance.

The paper provides a summary of the 2023 Unconstrained Ear Recognition Challenge (UERC), a benchmarking effort focused on ear recognition from images acquired in uncontrolled environments. The objective of the challenge was to evaluate the effectiveness of current ear recognition techniques on a challenging ear dataset while analyzing the techniques from two distinct aspects, i.e., verification performance and bias with respect to specific demographic factors, i.e., gender and ethnicity. Seven research groups participated in the challenge and submitted a seven distinct recognition approaches that ranged from descriptor-based methods and deep-learning models to ensemble techniques that relied on multiple data representations to maximize performance and minimize bias. A comprehensive investigation into the performance of the submitted models is presented, as well as an in-depth analysis of bias and associated performance differentials due to differences in gender and ethnicity. The results of the challenge suggest that a wide variety of models (e.g., transformers, convolutional neural networks, ensemble models) is capable of achieving competitive recognition results, but also that all of the models still exhibit considerable performance differentials with respect to both gender and ethnicity. To promote further development of unbiased and effective ear recognition models, the starter kit of UERC 2023 together with the baseline model, and training and test data is made available from: http://ears.fri.uni-lj.si/

Despite the significant progress made in pedestrian detection in last decade, detecting pedestrians under heavy occlusion still remains a challenging problem. In state of the art (SOTA), convolutional neural network (CNN) based models, the reason is attributed to non-maximal-suppression (NMS), which often erroneously deletes true positives when one pedestrian is occluding other. SOTA transformer based models do not have such NMS step, yet fail to detect highly occluded pedestrians. In this paper, we study the reasons for such failures. We observe that such models first predict key-points, and then compute the attention at the specific key-points. Our analysis reveals that the key-points do not have any preference towards semantically important body parts. Under heavy occlusion, such key-points end up attending to non-discriminative regions or background, leading to false negatives. We take inspiration from the conventional wisdom of detecting objects using their parts, and bias the attention of proposed transformer architecture towards semantically important, and highly discriminative human body parts. The intervention leads to SOTA results on benchmark Citypersons and Caltech datasets, achieving 30.75%, and 32.96% miss-rate (lower is better) respectively, against 32.6%, and 38.2% by the current SOTA. Code is available at https://ajayshastry08.github.io/pa_dino

A large proportion of videos captured today are first person videos shot from wearable cameras. Similar to other computer vision tasks, Deep Neural Networks (DNNs) are the workhorse for most state-of-the-art (SOTA) egocentric vision techniques. On the other hand DNNs are known to be susceptible to Adversarial Attacks (AAs) which add imperceptible noise to the input. Both black-box, as well as white-box attacks on image as well as video analysis tasks have been shown. We observe that most AA techniques basically add intensity perturbation to an image. Even for videos, the same process is essentially repeated for each frame independently. We note that definition of imperceptibility used for images may not be applicable for videos, where a small intensity change happening randomly in two consecutive frames may still be perceptible. In this paper we make a key novel suggestion to use perturbation in optical flow to carry out AAs on a video analysis system. Such perturbation is especially useful for egocentric videos, because there is lot of shake in the egocentric videos anyways, and adding a little more, keeps it highly imperceptible. In general our idea can be seen as adding structured, parametric noise as the adversarial perturbation. Our implementation of the idea by adding 3D rotations to the frames, reveal that using our technique, one can mount a black-box AA on an egocentric activity detection system in one-third of the queries compared to the SOTA AA technique.

In the COVID-19 situation, face masks have become an essential part of our daily life. As mask occludes most prominent facial characteristics, it brings new challenges to the existing facial recognition systems. This paper presents an idea to consider forehead creases (under surprise facial expression) as a new biometric modality to authenticate mask-wearing faces. The forehead biometrics utilizes the creases and textural skin patterns appearing due to voluntary contraction of the forehead region as features. The proposed framework is an efficient and generalizable deep learning framework for forehead recognition. Face-selfie images are collected using smartphone’s frontal camera in an unconstrained environment with various indoor/outdoor realistic environments. Acquired forehead images are first subjected to a segmentation model that results in rectangular Region Of Interest (ROI’s). A set of convolutional feature maps are subsequently obtained using a backbone network. The primary embeddings are enriched using a dual attention network (DANet) to induce discriminative feature learning. The attention-empowered embeddings are then optimized using Large Margin Co-sine Loss (LMCL) followed by Focal Loss to update weights for inducting robust training and better feature discriminating capabilities. Our system is end-to-end and few-shot; thus, it is very efficient in memory requirements and recognition rate. Besides, we present a forehead image dataset (BITS-IITMandi-ForeheadCreases Images Database1) that has been recorded in two sessions from 247 subjects containing a total of 4,964 selfie-face mask images. To the best of our knowledge, this is the first to date mobile-based fore-head dataset and is being made available along with the mobile application in the public domain. The proposed system has achieved high performance results in both closed-set, i.e., CRR of 99.08% and EER of 0.44% and open-set matching, i.e., CRR: 97.84%, EER: 12.40% which justifies the significance of using forehead as a biometric modality.

In egocentric videos, the face of a wearer capturing the video is never captured. This gives a false sense of security that the wearer’s privacy is preserved while sharing such videos. However, egocentric cameras are typically harnessed to wearer’s head, and hence, also capture wearer’s gait. Recent works have shown that wearer gait signatures can be extracted from egocentric videos, which can be used to determine if two egocentric videos have the same wearer. In a more damaging scenario, one can even recognize a wearer using hand gestures from egocentric videos, or identify a wearer in third person videos such as from a surveillance camera. We believe, this could be a death knell in sharing of egocentric videos, and fatal for egocentric vision research. In this work, we suggest a novel technique to anonymize egocentric videos, which create carefully crafted, but small, and imperceptible optical flow perturbations in an egocentric video’s frames. Importantly, these perturbations do not affect object detection or action/activity recognition from egocentric videos but are strong enough to dis-balance the gait recovery process. In our experiments on benchmark EPIC-Kitchens dataset, the proposed perturbation degrades the wearer recognition performance of [42], from 66.3% to 13.4%, while preserving the activity recognition performance of [10] from 89.6% to 87.4%. To test our anonymization with more wearer recognition techniques, we also developed a stronger, and more generalizable wearer recognition method based on camera egomotion cues. The approach achieves state-ofthe-art (SOTA) performance of 59.67% on EPIC-Kitchens, compared to 55.06% by [42]. However, the accuracy of our recognition technique also drops to 12% using the proposed anonymizing perturbations.

Biometric-based personal authentication systems have seen a strong demand mainly due to the increasing concern in various privacy and security applications. Although the use of each biometric trait is problem dependent, the human ear has been found to have enough discriminating characteristics to allow its use as a strong biometric measure. To locate an ear in a 2D side face image is a challenging task, numerous existing approaches have achieved significant performance, but the majority of studies are based on the constrained environment. However, ear biometrics possess a great level of difficulties in the unconstrained environment, where pose, scale, occlusion, illuminations, background clutter etc. varies to a great extent. To address the problem of ear localization in the wild, we have proposed two high-performance region of interest (ROI) segmentation models UESegNet-1 and UESegNet-2, which are fundamentally based on deep convolutional neural networks and primarily uses contextual information to localize ear in the unconstrained environment. Additionally, we have applied state-of-the-art deep learning models viz; FRCNN (Faster Region Proposal Network) and SSD (Single Shot MultiBox Detecor) for ear localization task. To test the model's generalization, they are evaluated on six different benchmark datasets viz; IITD, IITK, USTB-DB3, UND-E, UND-J2 and UBEAR, all of which contain challenging images. The performance of the models is compared on the basis of object detection performance measure parameters such as IOU (Intersection Over Union), Accuracy, Precision, Recall, and F1-Score. It has been observed that the proposed models UESegNet-1 and UESegNet-2 outperformed the FRCNN and SSD at higher values of IOUs i.e. an accuracy of 100\% is achieved at IOU 0.5 on majority of the databases.

Long acquisition time (AQT) due to series acquisition of multi-modality MR images (especially T2 weighted images (T2WI) with longer AQT), though beneficial for disease diagnosis, is practically undesirable. We propose a novel deep network based solution to reconstruct T2W images from T1W images (T1WI) using an encoder-decoder architecture. The proposed learning is aided with semantic features by using multi-channel input with intensity values and gradient of image in two orthogonal directions. A reconstruction module (RM) augmenting the network along with a domain adaptation module (DAM) which is an encoder-decoder model built-in with sharp bottleneck module (SBM) is trained via modular training. The proposed network significantly reduces the total AQT with negligible qualitative artifacts and quantitative loss (reconstructs one volume in (~1 second). The testing is done on publicly available dataset with real MR images, and the proposed network shows (~ 1dB) increase in PSNR over SOTA.

Massive biometric deployments are pervasive in today's world. But despite the high accuracy of biometric systems, their computational efficiency degrades drastically with an increase in the database size. Thus, it is essential to index them. Here, in this paper, we propose an iris indexing scheme using real-valued deep iris features binarized to iris bar codes (IBC) compatible with the indexing structure. Firstly, for extracting robust iris features, we have designed a network utilizing the domain knowledge of ordinal filtering and learning their nonlinear combinations. Later these real-valued features are binarized. Finally, for indexing the iris dataset, we have proposed a Mcomloss that can transform the binary feature into an improved feature compatible with the Multi-Index Hashing scheme. This Mcomloss function ensures the equal distribution of Hamming distance among all the contiguous disjoint sub-strings. To the best of our knowledge, this is the first work in the iris indexing domain that presents an end-to-end iris indexing structure. Experimental results on four datasets are presented to depict the efficacy of the proposed approach.

Learning Spatio-temporal features has shown improved performance on tasks involving video analysis using deep learning, and the deep learning community has used these features to solve a varied variety of problems. Video steganography is one such problem where learning these features for a video can help improve the performance of steganography. Steganography is the practice of concealing confidential information, to protect the information from an adversary, into an ordinary cover message in a way that the cover message does not seem suspicious to the adversary. Recent deep-learning-based steganography methods have proven to improve the secrecy and capacity of steganography over traditional techniques. In this paper, we propose a novel state-of-the-art deep 3D-CNN architecture with enhancement feature learning for full video steganography. The proposed model outperforms the current state-of-the-art methods for full video steganography both qualitatively and quantitatively. We have validated our model by comparing it with new as well as traditional steganography techniques, on quality and different statistical metrics, namely, PSNR, SSIM, APD, VIF at the frame, and video level. Moreover, to check the undetectability of our model, we have subjected our model to detection by steganalysis tools like SRNet. Results of fine-tuning classifiers, like ResNet and Inception-v3, to detect steganographic messages from ordinary messages maintains our model's undetectability and accuracy.

Attention Deficit Hyperactivity Disorder (ADHD) is a behavior-based disorder that mainly occurs in young children. Resting-state fMRI data have been very popular for diagnosing brain disorders like Autism, ADHD, and schizophrenia, by network-based functional connectivity, since these disorders are associated with both individual brain regions and their connections. Finding patterns among regions of controls' brain and ADHD patients' discriminating brains, is a non-trivial task. For classification of ADHD, we propose an end-to-end lightweight CNN architecture with hierarchical representation learning i.e., HLGSNet. We extract 116 anatomical regions from each subject in both normal and patient conditions, and graphs are built with the help of temporal correlation between different regions, where each region is considered as a node. Following this, a Siamese graph convolution neural network with triplet loss has been trained for finding embeddings so that samples for the same class should have similar embeddings. Finally, along with a fully connected layer, the trained model has been fine-tuned for the classification task. Experiments have been carried out on publicly available ADHD-200 dataset with promising performance.

Wearable egocentric cameras are typically harnessed to a wearer's head, giving them the unique advantage of capturing their points of view. Hoshen and Peleg have shown that egocentric cameras indirectly capture the wearer's gait, which can be used to identify a wearer based on their egocentric videos. The authors have shown a wearer recognition accuracy of up to 77% over 32 subjects. However, an important limitation of their work is that such gait features can be extracted only from walking sequences of a wearer. In this work, we take the privacy threat a notch higher and show that even the wearer's hand gestures, as seen through an egocentric video, leak wearer's identity. We have designed a model to extract and match hand gesture signatures from egocentric videos. We demonstrate the threat on the EPIC kitchen dataset containing 55 hours of the egocentric videos acquired from 32 subjects doing various activities. We show that: (1) Our model can recognize a wearer with an accuracy of up to 73% based on the same activity, i.e., the model has seen 'cut' activity by a wearer in the train set, and recognizes the wearer based on another 'cut' activity by him/her while testing. (2) The hand gesture signatures transfer across activities, i.e., even if our model does not see 'cut' activity of a wearer at the train time, but sees other activities such as 'wash', 'mix' etc., the model can still recognize a wearer with an accuracy of up to 60%, by matching hand gesture signatures of 'cut' at test time with train time signatures of 'wash' or 'mix'. (3) The hand gesture features even transfer across subjects, i.e., even if the model has not seen any activity by some subject, one can still verify a wearer (open-set), and predict that the same wearer has performed both activities with an Equal Error Rate of 15.21%. The code, trained models are available at https://egocentricbiometric.github.io/

We address the task of estimating depth from a single intensity image via a novel convolutional neural network (CNN) encoder-decoder architecture, which learns the depth information using example pairs of color images and their corresponding depth maps. The proposed model integrates residual connections within pooling and up-sampling layers, and hourglass networks which operate on the encoded features, thus processing these at various scales. Furthermore, the model is optimized under the constraints of perceptual as well as the mean squared error loss. The perceptual loss considers the high-level features, thus operating at a different scale of abstraction, which is complementary to the mean squared error loss. The improvements in qualitative and quantitative comparisons with state-of-the-art approaches demonstrate the effectiveness of our approach, even in presence of noise.

Recently, deep hierarchically learned models (such as CNN) have achieved superior performance in various computer vision tasks but limited attention has been paid to biometrics till now. This is major because of the number of samples available in biometrics are limited and are not enough to train CNN efficiently. However, deep learning often requires a lot of training data because of the huge number of parameters to be tuned by the learning algorithm. How about designing an end-to-end deep learning network to match the biometric features when the number of training samples is limited? To address this problem, we propose a new way to design an end-to-end deep neural network that works in two major steps: first an auto-encoder has been trained for learning domain specific features followed by a Siamese network trained via. triplet loss function for matching. A publicly available vein image data set has been utilized as a case study to justify our proposal. We observed that transformations learned from such a network provide domain specific and most discriminative vascular features. Subsequently, the corresponding traits are matched using multimodal pipelined end-to-end network in which the convolutional layers are pre-trained in an unsupervised fashion as an autoencoder. Thorough experimental studies suggest that the proposed framework consistently outperforms several state-of-the-art vein recognition approaches.

The aforementioned works and other analogous studies in finger knuckle images recognition have claimed that the precise detection of true features is difficult from poorly segmented images and the main reason for matching errors. Thus, an accurate segmentation of the region of interest is very crucial to achieve superior recognition results. In this paper, we have proposed a novel holistic and generalized segmentation Network (HFDSegNet) that automatically categorizes the given finger dorsal image obtained from multiple sensory resources into particular class and then extracts three possible ROIs (major knuckle, minor knuckle and nail) accurately. To best of our knowledge, this is the first attempt, an end-to-end trained object detector inspired by Deep Learning technique namely faster R-CNN (Region based Convolutional Neural Network) has been employed to detect and localize the position of finger knuckles and nail, even finger images exhibit blur, occlusion, low contrast etc. The experimental results are examined on two publicly available databases named as Poly-U contact-less FKI data-set, and Poly U FKP database. The proposed network is trained only over 500 randomly selected images per database, demonstrate the outstanding performance of proposed ROI’s segmentation network.

Current finger knuckle image recognition systems, often require users to place fingers' major or minor joints flatly towards the capturing sensor. To extend these systems for user non-intrusive application scenarios, such as consumer electronics, forensic, defence etc, we suggest matching the full dorsal fingers, rather than the major/ minor region of interest (ROI) alone. In particular, this paper makes a comprehensive study on the comparisons between full finger and fusion of finger ROI's for finger knuckle image recognition. These experiments suggest that using full-finger, provides a more elegant solution. Addressing the finger matching problem, we propose a CNN (convolutional neural network) which creates a 128-D feature embedding of an image. It is trained via. triplet loss function, which enforces the L2 distance between the embeddings of the same subject to be approaching zero, whereas the distance between any 2 embeddings of different subjects to be at least a margin. For precise training of the network, we use dynamic adaptive margin, data augmentation, and hard negative mining. In distinguished experiments, the individual performance of finger, as well as weighted sum score level fusion of major knuckle, minor knuckle, and nail modalities have been computed, justifying our assumption to consider full finger as biometrics instead of its counterparts. The proposed method is evaluated using two publicly available finger knuckle image datasets i.e., PolyU FKP dataset and PolyU Contactless FKI Datasets.

We propose an approach to reconstruct high field (7T) like MR images from low field (3T) MR images, which involves a merged convolutional autoencoder neural network. The network uses merge connections from downsampling encoder layers to cascade the input at upsampling layers of the decoder in order to preserve the local image details in transformed space. Further, we have used three channel input, defined by its image intensity values and corresponding gradient values in order to guide the discrimination. In terms of comparison with state-of-the-art, the proposed algorithm reconstruct 7T MR images with better tissue contrast, yields quantitative improvements, and has a significantly more efficient run-time. We also demonstrate the effectiveness of the approach with low training data and noise.

Cloud environment provides on-demand access to a shared pool of computing resources over the Internet. Failures are unavoidable in such a distributed and complex environment. In this work, we simulate such a scenario in a docker based virtual environment to aid a proactive approach for anomaly identification in a cloud environment. Proactive approach involves resource prediction first and then anomaly detection. This paper focuses only on resource prediction. We also propose a hybrid model of LSTM and BLSTM using association learning that captures the relationship between the related resource metrics to predict future resource workload in cloud. We use a mix of different types of workloads for simulating the workloads in a cloud environment. The proposed approach is validated on the collected trace of data in a docker based virtual environment as well as the Google cluster trace. It is observed that the proposed model works better as compared to the other state-of-the-art models for resource workload prediction.

Human identification has now been a social liability due to frequent terror threats and corrupt bureaucratic practices, especially in rural countries like India. It has been surprisingly observed that fingerprint quality is poor as compared with finger knuckle quality of rural users as they exist on the outer hand side. In this paper, we are proposing a novel finger-knuckle-print based identification system. Initially, finger knuckle image is pre-processed using proposed local and adaptive image transformations. Then, finger knuckle image matching has been performed using multi-scale Deep-Matching technique. Finally, score level fusion rule has been employed to achieve competitive performance over public FKP database.

Iris is considered as one of the best biometric trait for human authentication due to its accuracy and permanence. However easy iris spoofing raise the risk of false acceptance or false rejection. Recent iris recognition research has made an attempt to quantify the performance degradation due to the use of contact lens. This study proposes a strategy to detect soft contact lens in visual pictures of the eye obtained using NIR sensor. The lens border is detected by considering small annular ring-like area near the outer iris boundary and locating candidate points while traversing along the lens perimeter. The system performance is evaluated over public databases such as IIITD-Cogent, UND 2010, IIITD-Vista along with our self created IITK database. The rigorous experimentation revels the superior performance of the proposed system as compared with other existing techniques.

Iris biometrics is widely used because of highly discriminative characteristics that are found in iris. But designing an iris recognition system which is invariant to extrinsic factors such as illumination, noise, camera-to-eye distance is fairly challenging. This paper proposes a novel iris recognition approach which takes into account the iris structure, illumination variation, occlusion, noise and rotational variance. A feature extraction technique exploiting the local iris features has been proposed that uses two features extracted from different blocks of an iris. The features are extracted from suitably modified pixel values which are enhanced to improve robustness of the technique. Relational measure is proposed that considers both radial and circumferential features which is combined with the block local binary pattern (BLBP). The BLBP is applied in an unconventional block-wise manner adaptive to the iris structure. Finally scores are fused at score level. Experimental results on two publicly available Casia Interval and Lamp databases as well as on our IITK iris database demonstrates the usefulness of the proposed system.

Image quality can play key role in the system performance. The recently introduced knuckleprint biometric has shown promising results, but its quality assessment is difficult because it lacks well defined and structured features as in the case of face or fingerprint. To our knowledge this is the first attempt to automatically assess the quality of knuckleprint images. The quality of knuckleprint images mainly depends upon the vertical line like features, focus, contrast and reflections produced by the camera flash. In this paper an effort has been made to identify, estimate and quantify some of these quality attributes and fuse them to obtain an overall quality score for any knuckleprint image. The largest publicly available PolyU knuckleprint database is used for testing, containing 7920 images. Extensive study is being carried out in order to establish a relationship between image quality and matching performance in order to demonstrate the proposed framework's utility.

This paper proposes a dissimilarity measure that can be used as a distance between two images. It has shown better discriminative power to recognize faces than similar existing variants for discriminating facial images. It gives more weight to the pixels which are often a part of the edge and counts pixels that are unmatched between query and database images. This measure has been tested on a publicly available database as ORL, YALE, CALTEC, BERN and also on a database developed at IIT Kanpur. Experimental results show that the proposed measure achieves a high recognition rate of 99.75% 93.75% 99.03% 98.93% 99.73% for the first likely matched faces on databases ORL, YALE, BERN, CALTECH, IITK respectively. The proposed measure can provide not only effective result against pose and expression variations but also against slight illumination variation.

This paper proposes a new powerful distance measure called Normalized Unmatched Points (NUP). This measure can be used in a face recognition system to discriminate facial images. It works by counting the number of unmatched pixels between query and database images. A face recognition system has been proposed which makes use of this proposed distance measure for taking the decision on matching. This system has been tested on four publicly available databases, viz. ORL, YALE, BERN and CALTECH databases. Experimental results show that the proposed measure achieves recognition rates more than 98.66% for the first five likely matched faces. It is observed that the NUP distance measure performs better than other existing similar variants on these databases.