A5053653138- Biometric Identification and Security
- Face recognition and analysis
- User Authentication and Security Systems
- COVID-19 diagnosis using AI
- Radiomics and Machine Learning in Medical Imaging
- Data Quality and Management
- Forensic Fingerprint Detection Methods
- Text and Document Classification Technologies
- Advanced Graph Neural Networks
- Digital Media Forensic Detection
- Lung Cancer Diagnosis and Treatment
- Anomaly Detection Techniques and Applications
- Topic Modeling
- Web Data Mining and Analysis
- Gait Recognition and Analysis
- Face and Expression Recognition
- Wood and Agarwood Research
- Domain Adaptation and Few-Shot Learning
- Vehicle License Plate Recognition
- Rough Sets and Fuzzy Logic
- Electricity Theft Detection Techniques
- Artificial Intelligence in Healthcare
- 3D Shape Modeling and Analysis
- Advanced Image and Video Retrieval Techniques
- AI in cancer detection
Mastercard (United States)
2024
Indian Institute of Technology Delhi
2015-2023
Indraprastha Institute of Information Technology Delhi
2012-2023
Indian Institute of Technology Jodhpur
2022
With the growing popularity and usage of online social media services, people now have accounts (some times several) on multiple diverse services like Facebook, Linked In, Twitter You Tube. Publicly available information can be used to create a digital footprint any user using these services. Generating such footprints very useful for personalization, profile management, detecting malicious behavior users. A important application analyzing users' is protect users from potential privacy...
Authenticating fingerphoto images captured using a smartphone camera, provide good alternate solution in place of traditional pin or pattern based approaches. There are multiple challenges associated with authentication such as background variations, environmental illumination, estimating finger position, and camera resolution. In this research, we propose novel ScatNet feature matching approach. Effective segmentation enhancement performed to aid the process attenuate effect capture...
With the advancements in technology, smartphones' capabilities have increased immensely. For instance, smartphone cameras are being used for face and ocular biometric-based authentication. This research proposes finger-selfie based authentication mechanism, which uses a camera to acquire selfie of finger. In addition personal device-level authentication, finger-selfies may also be matched with livescan fingerprints present legacy/national ID databases remote or touchless We propose an...
In many applications such as law enforcement, attendance systems, and medical services, biometrics is utilized for identifying individuals. However, current in general, do not enroll all possible age groups, particularly, toddlers pre-school children. This research the first of its kind attempt to prepare a multimodal biometric database potential users systems. proposed database, face, fingerprint, iris modalities over 100 children (age range 18 months 4 years) are captured two different...
Biometrics based user authentication for mobile devices is now popular with face and fingerprints being the primary modalities. Fingerphoto, an image of a person's finger captured using inbuilt smartphone camera, attractive cost-effective alternative. Existing research focuses on constrained or semi-constrained environment; whereas, challenges such as cooperation, number fingers, background, orientation, deformation are important to address before fingerphoto becomes usable. This paper...
Entity Alignment (EA) is the task of recognizing same entity present in different knowledge bases. Recently, embedding-based EA techniques have established dominance where alignment done based on closeness latent space. Graph Neural Networks (GNN) gained popularity as embedding module due to its ability learn entities' representation their local sub-graph structures. Although GNN shows promising results, limited works aimed capture relations while considering global importance and relative...
Latent fingerprint recognition involves acquisition and comparison of latent fingerprints with an exemplar gallery fingerprints. The diversity in the type surface leads to different procedures recover fingerprint. appearance vary significantly due development techniques, leading large intra-class variation. Due lack datasets acquired using multiple mechanisms surfaces, existing algorithms for enhancement may perform poorly. In this study, we propose a Multi-Surface Multi-Technique (MUST)...
The latent fingerprint examiners often mark minutiae and perform a comparison of fingerprints with exemplar known identity. interpretation details in the is based on proficiency examiners. Different discern regions differently due to an unconscious choice certain features driven by their experiences. In this study, we aim draw inferences from perceptual behavior collecting eye gaze while they comparison. study shows patterns observed across different forensic infers specific heuristics used...
With the availability of smartphone cameras, high speed internet, and connectivity to social media, users post content on go including check-ins, text, images. Privacy leaks due posts related check-ins text is an issue in itself, however, this paper discusses potential leak one's biometric information via images posted media. While posting photos themselves or highlighting miniature objects, end up that leads irreversible loss such as ocular region, fingerprint, knuckle print, ear print. In...
Consistent clinical observations of characteristic findings COVID-19 pneumonia on chest X-rays have attracted the research community to strive provide a fast and reliable method for screening suspected patients. Several machine learning algorithms been proposed find abnormalities in lungs using specific distinguish them from other etiologies pneumonia. However, despite enormous magnitude pandemic, there are very few instances public databases pneumonia, best our knowledge, is no database...
Abstract Purpose To advance the usage of CXRs as a viable solution for efficient COVID-19 diagnostics by providing large-scale annotations abnormalities in frontal BIMCV-COVID19+ database, and to provide robust evaluation mechanism facilitate its usage. Materials Methods We abnormality creating bounding boxes. The are part existing database. also define four different protocols semantic segmentation classification algorithms. Finally, we benchmark defined report results using popular deep...
Biometric authentication during the COVID-19 and post-pandemic times require a touchless mechanism. While existing studies showcase use of fingerphoto for authentication, short video finger can provide many good-quality frames. This research presents first publicly available finger-video dataset, titled Multi-Movement Finger-Video (MMFV) Database. The MMFV dataset has 3792 videos from 336 classes, acquired over two sessions, spans three different movement types (pitch, yaw, roll). To...
Temporal Point Processes (TPP) are probabilistic generative frameworks. They model discrete event sequences localized in continuous time. Generally, real-life events reveal descriptive information, known as marks. Marked TPPs time and marks of the together for practical relevance. Conditioned on past events, marked aim to learn joint distribution mark next event. For simplicity, conditionally independent TPP models assume given history. factorize conditional into product individual...
Scalability and training time are crucial for any graph neural network model processing a knowledge (KG). While partitioning graphs helps reduce the time, prediction accuracy reduces significantly compared to on whole graph. In this paper, we propose CPa-WAC: lightweight architecture that incorporates convolutional networks modularity maximization-based constellation harness power of local topology. The proposed CPa-WAC method memory cost embedding, making learning scalable. results from our...
Classification tasks present challenges due to class imbalances and evolving data distributions. Addressing these issues requires a robust method handle while effectively detecting out-of-distribution (OOD) samples not encountered during training. This study introduces novel OOD detection algorithm designed for tabular datasets, titled Deep Neural Network-based Gaussian Descriptor Imbalanced Tabular Data (DNN-GDITD). The DNN-GDITD can be placed on top of any DNN facilitate better...
Many existing Graph Neural Networks (GNN) methods assume that labels are reliable and sufficient, which may not be the case in real-world scenarios. This paper addresses one such problem of Partial Label Learning (PLL) on graph-structured data. In PLL for graphs, each node is represented by a candidate set labels, where only true while others inaccurate. Despite advancements with tabular vision domains, data still needs to explored. this work, we first define graphs. Subsequently, propose...
Entity Alignment (EA) identifies entities across databases that refer to the same entity. Knowledge graph-based embedding methods have recently dominated EA techniques. Such map a low-dimension space and align them based on their similarities. With corpus of methodologies growing rapidly, this paper presents comprehensive analysis various existing methods, elaborating applications limitations. Further, we distinguish underlying algorithms information they incorporate learn entity...
In many surveillance applications, the cameras are placed at overhead heights for human identification. such real-world scenarios, person of interest might be walking away from camera and only information available is "image person's head". this research, we investigate usage head images recognition propose it as a soft-biometric modality. With its viability recognition, application can also extended with other face algorithms surveillance. We image database pertaining to 103 subjects more...