arXiv subjects

Rajeev Ranjan

Publications and source records attributed to Rajeev Ranjan.

At least 37 records · Page 2Linked to original sources

arXiv2018

Light-weight Head Pose Invariant Gaze Tracking

Unconstrained remote gaze tracking using off-the-shelf cameras is a challenging problem. Recently, promising algorithms for appearance-based gaze estimation using convolutional neural networks (CNN) have been proposed. Improving their robustness to various confounding factors including variable head pose, subject identity, illumination and image quality remain open problems. In this work, we study the effect of variable head pose on machine learning regressors trained to estimate gaze direction. We propose a novel branched CNN architecture that improves the robustness of gaze classifiers to variable head pose, without increasing computational cost. We also present various procedures to effectively train our gaze network including transfer learning from the more closely related task of object viewpoint estimation and from a large high-fidelity synthetic gaze dataset, which enable our ten times faster gaze network to achieve competitive accuracy to its current state-of-the-art direct competitor.

cs.CV↗

arXiv2018

Improving Network Robustness against Adversarial Attacks with Compact Convolution

Though Convolutional Neural Networks (CNNs) have surpassed human-level performance on tasks such as object classification and face verification, they can easily be fooled by adversarial attacks. These attacks add a small perturbation to the input image that causes the network to misclassify the sample. In this paper, we focus on neutralizing adversarial attacks by compact feature learning. In particular, we show that learning features in a closed and bounded space improves the robustness of the network. We explore the effect of L2-Softmax Loss, that enforces compactness in the learned features, thus resulting in enhanced robustness to adversarial perturbations. Additionally, we propose compact convolution, a novel method of convolution that when incorporated in conventional CNNs improves their robustness. Compact convolution ensures feature compactness at every layer such that they are bounded and close to each other. Extensive experiments show that Compact Convolutional Networks (CCNs) neutralize multiple types of attacks, and perform better than existing methods in defending adversarial attacks, without incurring any additional training overhead compared to CNNs.

cs.LG↗

arXiv2018

Length-scale dependent average structures, piezoelectricity enhancement and depolarization mechanisms in a non-MPB high-performance piezoelectric alloy system PbTiO3-Bi(Zr1/2Ni1/2)O3

There is a general perception that large piezoelectric response in ferroelectric alloys requires tuning the system towards a morphotropic phase boundary (MPB), i.e., a composition driven inter-ferroelectric instability. Here we show that high piezoelectric response can be realized even in non-MPB alloy systems. This is demonstrated on (1-x)PbTiO3-(x)Bi(Zr0.5Ni0.5)O3 (PT-BNZ) by a comprehensive study involving electric-field and temperature dependent XRD, Raman spectroscopy, dielectric, piezoelectric and high field electrostrain measurements. We found that poling-field irreversibly suppresses the cubic-like phase at room temperature. Based on our results, we argue that that which appears as MPB, comprising of tetragonal and cubic-like phases on the global scale, is not so actually. The large piezoresponse is due to coexistence of tetragonal regions of long and short-range coherence. The PT-BNZ system is therefore qualitatively different from the conventional MPB systems such as PZT, PMN-PT, and PbTiO3-BiScO3, etc., which exhibits coexisting tetragonal and rhombohedral/monoclinic phases in thermodynamic equilibrium. In the absence of inter-ferroelectric instability as a phenomenon, field induced polarization-rotation and inter-ferroelectric transformation are no longer plausible mechanisms to explain the large piezoelectric response in PT-BNZ. The large piezoelectricity is primarily due to enhanced mobility of the tetragonal domain walls enabled by domain miniaturization. Our study proves that attainment of large piezoelectricity does not require inter-ferroelectric instability as a necessary criterion.

cond-mat.mtrl-sci↗

arXiv2017

HyperFace: A Deep Multi-task Learning Framework for Face Detection, Landmark Localization, Pose Estimation, and Gender Recognition

We present an algorithm for simultaneous face detection, landmarks localization, pose estimation and gender recognition using deep convolutional neural networks (CNN). The proposed method called, HyperFace, fuses the intermediate layers of a deep CNN using a separate CNN followed by a multi-task learning algorithm that operates on the fused features. It exploits the synergy among the tasks which boosts up their individual performances. Additionally, we propose two variants of HyperFace: (1) HyperFace-ResNet that builds on the ResNet-101 model and achieves significant improvement in performance, and (2) Fast-HyperFace that uses a high recall fast face detector for generating region proposals to improve the speed of the algorithm. Extensive experiments show that the proposed models are able to capture both global and local information in faces and performs significantly better than many competitive algorithms for each of these four tasks.

cs.CV↗

arXiv2017

Grain-size dependent electric-field induced structural changes in relaxor-ferroelectric based unclamped piezoelectric grains

Polymer-piezoceramic 0-3 composites combine the flexibility of the polymers and the excellent piezoelectric properties of the ferroelectric based ceramic. While grain size of the ceramic powder is one of the important considerations in the fabrication of such composites, a correlation relating poling field induced structural changes and its possible influence on the overall piezoelectric response of the composite is still lacking. In this paper, we examine this issue on a 0-3 piezo-composite comprising of a ceramic powders of a low-lead piezoelectric alloy (x)Bi(Ni1/2Zr1/2O3-(1-x)PbTiO3 in proximity of its morphotropic phase boundary, and polyvinylidene fluoride (PVDF) as the polymer component. Composites were fabricated by fixing the volume fraction of the ceramic while varying the grain size. We found a non-monotonic variation in the piezo-response as a function of grain size. Structural analysis before and after poling of the piezo-composites revealed evidence of poling induced cubic-like to tetragonal irreversible transformation, the extent of which is dependent on the grain size.

cond-mat.mtrl-sci↗

arXiv2017

Correlation between structure and Rayleigh parameters in the lead-free piezoceramic (1-x)Ba(Ti0.88 Sn0.12)O3-x(Ba0.7Ca0.3)TiO3

Composition dependent Rayleigh and structural analysis was carried out on the lead-free piezoceramics (1-x)(BaTi0.88Sn0.12)-x(Ba0.7Ca0.3)TiO3 at room temperature. The system exhibits tetragonal (P4mm) structure for x > 0.21, rhombohedral (R3m) for x < 0.13 and orthorhombic (Amm2) for 0.13<x<0.21. Rayleigh analysis suggests that the irreversible contribution to the dielectric response is enhanced in the single phase orthorhombic compositions in the vicinity of the R3m-Amm2 and Amm2-P4mm phase boundaries, and not in compositions exhibiting phase coexistences (x = 0.12 and 0.22). We also found a correspondence between the irreversible Rayleigh parameter and the coercive field in this system.

cond-mat.mtrl-sci↗

arXiv2017

Tuning photoluminescence response by electric field in the lead-free piezoelectric Na1/2Bi1/2TiO3-BaTiO3

We show that an electrically soft ferroelectric host can be used to tune the photoluminescence (PL) response of rare-earth emitter ions by external electric field. The proof of this concept is demonstrated by changing the PL response of Eu+3 ion by electric field on a model system Eu-doped 0.94Na1/2Bi1/2TiO3-0.06BaTiO3. We also show that new channels of radiative transitions, forbidden otherwise, open up due to positional disorder in the system, which can as well be tuned by electric field.

cond-mat.mtrl-sci↗

arXiv2017

A structural perspective on the origin of the anomalous weak-field piezoelectric response at the polymorphic phase boundaries of (Ba, Ca)(Ti, M)O3 lead-free piezoelectrics

In this paper, we compare the structural features of Ca modified Ba(Ti, Sn)O3 with that of its Ca-free counterpart to understand why the Ca-modified variant of this lead-free piezoelectric gives a much higher piezoelectric response. We found that in spite of the reduced spontaneous lattice strain, the polarization of the Ca-modified variant is higher. We, therefore, demonstrate that the very large piezoelectric response of (Ba, Ca)(Ti, Sn)O3 is associated with the decoupling of the polarization and lattice strain by Ca-modification.

cond-mat.mtrl-sci↗

arXiv2017

Magnetic field driven enhanced ferroelectric switching in self-grown ferroelectric-ferromagnetic composite in the BiFeO3-BaTiO3 multiferroic alloy system

Over the years attempts have been made to compensate for the inherent weaknesses in the bulk state of the multiferroic BiFeO3, such as high leakage current and the absence of ferromagnetic correlation, and exploit its magnetoelectric potential by forming solid solutions with other perovskites. Studies in the recent few years have shown that alloying of BiFeO3 with BaTiO3, both with and without additives, can induce both ferroelectric and ferromagnetic switching. While the coexistence of both the ferroic orders is encouraging from the view point of technological applications, the origin of ferromagnetism in this system remains elusive. Here, we synthesized such compositions and carried out a detailed structural analysis employing magnetic separation of the powder particles. We found that the origin of ferromagnetism lies in the spontaneous precipitation of a minor ( ~ 1 wt %) barium hexaferrite (BaFe9O19) phase, often undetected in routine x-ray diffraction studies of powders sampled from the entire specimen. We also demonstrate that inspite of the insignificant fraction the ferrimagnetic phase, this self-grown composite exhibit noticeably enhanced ferroelectric switching in the presence of external magnetic field. We obtained a dc magnetoelectric coupling of ~ 9 x 10-8 s/m, a value which is comparable to what has been reported for layered ferroelectric/ferromagnetic laminates and bilayer thin film ferroelectric-ferromagnetic hetrostructures. Our study suggests that reasonably large magnetoelectric coupling is realizable in simple 0-3 ferroelectric-ferromagnetic bulk composites provided synthesis strategies are developed which induces spontaneous precipitation of the ferromagnetic phase in small volume fraction to ensure good insulating behaviour of the composite thus developed.

cond-mat.mtrl-sci↗

arXiv2017

The Do's and Don'ts for CNN-based Face Verification

While the research community appears to have developed a consensus on the methods of acquiring annotated data, design and training of CNNs, many questions still remain to be answered. In this paper, we explore the following questions that are critical to face recognition research: (i) Can we train on still images and expect the systems to work on videos? (ii) Are deeper datasets better than wider datasets? (iii) Does adding label noise lead to improvement in performance of deep networks? (iv) Is alignment needed for face recognition? We address these questions by training CNNs using CASIA-WebFace, UMDFaces, and a new video dataset and testing on YouTube- Faces, IJB-A and a disjoint portion of UMDFaces datasets. Our new data set, which will be made publicly available, has 22,075 videos and 3,735,476 human annotated frames extracted from them.

cs.CV↗

arXiv2017

Unconstrained Still/Video-Based Face Verification with Deep Convolutional Neural Networks

Over the last five years, methods based on Deep Convolutional Neural Networks (DCNNs) have shown impressive performance improvements for object detection and recognition problems. This has been made possible due to the availability of large annotated datasets, a better understanding of the non-linear mapping between input images and class labels as well as the affordability of GPUs. In this paper, we present the design details of a deep learning system for unconstrained face recognition, including modules for face detection, association, alignment and face verification. The quantitative performance evaluation is conducted using the IARPA Janus Benchmark A (IJB-A), the JANUS Challenge Set 2 (JANUS CS2), and the LFW dataset. The IJB-A dataset includes real-world unconstrained faces of 500 subjects with significant pose and illumination variations which are much harder than the Labeled Faces in the Wild (LFW) and Youtube Face (YTF) datasets. JANUS CS2 is the extended version of IJB-A which contains not only all the images/frames of IJB-A but also includes the original videos for evaluating the video-based face verification system. Some open issues regarding DCNNs for face verification problems are then discussed.

cs.CV↗

arXiv2017

L2-constrained Softmax Loss for Discriminative Face Verification

In recent years, the performance of face verification systems has significantly improved using deep convolutional neural networks (DCNNs). A typical pipeline for face verification includes training a deep network for subject classification with softmax loss, using the penultimate layer output as the feature descriptor, and generating a cosine similarity score given a pair of face images. The softmax loss function does not optimize the features to have higher similarity score for positive pairs and lower similarity score for negative pairs, which leads to a performance gap. In this paper, we add an L2-constraint to the feature descriptors which restricts them to lie on a hypersphere of a fixed radius. This module can be easily implemented using existing deep learning frameworks. We show that integrating this simple step in the training pipeline significantly boosts the performance of face verification. Specifically, we achieve state-of-the-art results on the challenging IJB-A dataset, achieving True Accept Rate of 0.909 at False Accept Rate 0.0001 on the face verification protocol. Additionally, we achieve state-of-the-art performance on LFW dataset with an accuracy of 99.78%, and competing performance on YTF dataset with accuracy of 96.08%.

cs.CV↗

arXiv2017

UMDFaces: An Annotated Face Dataset for Training Deep Networks

Recent progress in face detection (including keypoint detection), and recognition is mainly being driven by (i) deeper convolutional neural network architectures, and (ii) larger datasets. However, most of the large datasets are maintained by private companies and are not publicly available. The academic computer vision community needs larger and more varied datasets to make further progress. In this paper we introduce a new face dataset, called UMDFaces, which has 367,888 annotated faces of 8,277 subjects. We also introduce a new face recognition evaluation protocol which will help advance the state-of-the-art in this area. We discuss how a large dataset can be collected and annotated using human annotators and deep networks. We provide human curated bounding boxes for faces. We also provide estimated pose (roll, pitch and yaw), locations of twenty-one key-points and gender information generated by a pre-trained neural network. In addition, the quality of keypoint annotations has been verified by humans for about 115,000 images. Finally, we compare the quality of the dataset with other publicly available face datasets at similar scales.

cs.CV↗

arXiv2016

An All-In-One Convolutional Neural Network for Face Analysis

We present a multi-purpose algorithm for simultaneous face detection, face alignment, pose estimation, gender recognition, smile detection, age estimation and face recognition using a single deep convolutional neural network (CNN). The proposed method employs a multi-task learning framework that regularizes the shared parameters of CNN and builds a synergy among different domains and tasks. Extensive experiments show that the network has a better understanding of face and achieves state-of-the-art result for most of these tasks.

cs.CV↗

arXiv2016

Face Alignment by Local Deep Descriptor Regression

We present an algorithm for extracting key-point descriptors using deep convolutional neural networks (CNN). Unlike many existing deep CNNs, our model computes local features around a given point in an image. We also present a face alignment algorithm based on regression using these local descriptors. The proposed method called Local Deep Descriptor Regression (LDDR) is able to localize face landmarks of varying sizes, poses and occlusions with high accuracy. Deep Descriptors presented in this paper are able to uniquely and efficiently describe every pixel in the image and therefore can potentially replace traditional descriptors such as SIFT and HOG. Extensive evaluations on five publicly available unconstrained face alignment datasets show that our deep descriptor network is able to capture strong local features around a given landmark and performs significantly better than many competitive and state-of-the-art face alignment algorithms.

cs.CV↗

arXiv2016

Towards the Design of an End-to-End Automated System for Image and Video-based Recognition

Over many decades, researchers working in object recognition have longed for an end-to-end automated system that will simply accept 2D or 3D image or videos as inputs and output the labels of objects in the input data. Computer vision methods that use representations derived based on geometric, radiometric and neural considerations and statistical and structural matchers and artificial neural network-based methods where a multi-layer network learns the mapping from inputs to class labels have provided competing approaches for image recognition problems. Over the last four years, methods based on Deep Convolutional Neural Networks (DCNNs) have shown impressive performance improvements on object detection/recognition challenge problems. This has been made possible due to the availability of large annotated data, a better understanding of the non-linear mapping between image and class labels as well as the affordability of GPUs. In this paper, we present a brief history of developments in computer vision and artificial neural networks over the last forty years for the problem of image-based recognition. We then present the design details of a deep learning system for end-to-end unconstrained face verification/recognition. Some open issues regarding DCNNs for object recognition problems are then discussed. We caution the readers that the views expressed in this paper are from the authors and authors only!

cs.CV↗

arXiv2015

Extrinsic Rayleigh coefficient and domain mobility in critical MPB compositions of high-performance piezoceramics: A revisit

Enhanced domain mobility is considered as one of the important characteristics of morphotropic phase boundary (MPB) ferroelectrics exhibiting a very large piezoelectric response. For nearly two decades, Rayleigh analysis of dielectric and piezoelectric response at sub-coercive fields has been used as a phenomenological tool to determine the relative contribution of domain walls in influencing polar properties in these systems. The high value of the extrinsic Rayleigh coefficient at the MPB is generally attributed to enhanced domain mobility. Using coercive field as a measure of hindrance to domain mobility, in this paper, we have examined the correlation between the extrinsic Rayleigh coefficient, piezoelectric response, and coercive field (EC) by a comparative study on three different high performance piezoelectrics (i) (BaTi0.88Sn0.12)-(Ba0.7Ca0.3)TiO3, (ii) Pb(Zr,Ti)O3 and (iii) PbTiO3-BiScO3 across their respective MPBs. Our study shows that although the three systems exhibit maximum piezoelectric response and maximum extrinsic Rayleigh coefficient for their respective MPB compositions, the domain mobility in the MPB compositions of the Pb-based systems is not the highest. The lead-free system, on the other hand, exhibits highest domain mobility for the critical MPB composition. Contrary to the common perception, our study reveals that large value of the Rayleigh coefficient at the MPB is not necessarily indicative of enhanced domain mobility. We argue that the large piezoresponse in MPB systems is a result of complex interplay of field induced motion of the interphase boundaries and domain walls. In the lead-based systems the interphase boundary motion hinders domain wall mobility. The large value of the Rayleigh coefficient and piezoresponse in the Pb-based systems is because of the dominant role of the interphase boundary motion instead of the enhanced domain wall mobility.

cond-mat.mtrl-sci↗

arXiv2015

Evidence of field induced inter-ferroelectric transformation as the dominant driving mechanism for anomalous piezoresponse in the morphotropic phase boundary piezoelectric system PbTiO3-BiScO3

The contributory mechanisms associated with high piezoelectric response in piezoelectric ceramics have been examined by in-situ electric field dependent high energy synchro x-ray diffraction study. A comparative study of electric field induced lattice strain and the propensity for non-180o domain switching on two closeby compositions of a high performance piezoelectric alloy (1-x)PbTiO3-(x)BiScO3, one within the morphotropic phase boundary (MPB) region exhibiting d33 of 425 pC/N and another just outside the MPB region exhibiting d33 of 260 pC/N, unravelled that, inspite of the MPB specimen exhibiting considerably high piezoelectric response, its lattice strain and domain switching propensity is considerably less as compared to the non-MPB specimen. These new experimental observations contradict the commonly held view that anomalous piezoelectric response in MPB based piezoelectrics arise due to enhanced propensity for domain switching. Our results show the dominant mechanism contributing to the anomalous piezo-response is field induced interferroelectric transformation.

cond-mat.mtrl-sci↗