Swapna Agarwal - Academia.edu (original) (raw)
Papers by Swapna Agarwal
Noise & Vibration Worldwide
With increasing awareness of the need for quieter operation of industrial machines, much effort h... more With increasing awareness of the need for quieter operation of industrial machines, much effort has been put into reducing the emitted noise levels in their design, development and operation stages. This study characterizes emitted noise from Jhiri machines as well as suggesting suitable control measures to be incorporated for noise abatement.
Proceedings of the 18th Conference on Embedded Networked Sensor Systems
In this paper, we present "ThermoTrak", a smartphone accessory based, real-time and acc... more In this paper, we present "ThermoTrak", a smartphone accessory based, real-time and accurate temperature measurement mechanism, which can be used to screen for fever, which is a manifestation of infectious diseases including the symptoms caused by SARS-CoV-2. Our system accurately identifies face and forehead region from a safe distance of one meter, calculates accurate temperature of forehead with accuracy of ±0.5° C on a linear scale. An AI based algorithm is employed for the purpose of accurately detecting the Region of interest (ROI) (Face & point near center of Forehead) and calculate the absolute temperature within 300 milliseconds.
The paper propose a cognitive inspired change detection method for the detection and localization... more The paper propose a cognitive inspired change detection method for the detection and localization of shape variations on point clouds. A well defined pipeline is introduced by proposing a coarse to fine approach: i) shape segmentation, ii) fine segment registration using attention blocks. Shape segmentation is obtained using covariance based method and fine segment registration is carried out using gravitational registration algorithm. In particular the introduction of this partition-based approach using visual attention mechanism improves the speed of deformation detection and localization. Some results are shown on synthetic data of house and aircraft models. Experimental results shows that this simple yet effective approach designed with an eye to scalability can detect and localize the deformation in a faster manner. A real world car use case is also presented with some preliminary promising results useful for auditing and insurance claim tasks.
2017 IEEE International Conference on Image Processing (ICIP), Sep 1, 2017
In this paper, a shape-constrained iterative algorithm is proposed to register a rigid template p... more In this paper, a shape-constrained iterative algorithm is proposed to register a rigid template point-cloud to a given reference point-cloud. The algorithm embeds a shape-based similarity constraint into the principle of gravitation. The shape-constrained gravitation, as induced by the reference, controls the movement of the template such that at each iteration, the template better aligns with the reference in terms of shape. This constraint enables the alignment in difficult conditions indtroduced by change (presence of outliers and/or missing parts), translation, rotation and scaling. We discuss efficient implementation techniques with least manual intervention. The registration is shown to be useful for change detection in the 3D point-cloud. The algorithm is compared with three state-of-the-art registration approaches. The experiments are done on both synthetic and real-world data. The proposed algorithm is shown to perform better in the presence of big rotation, structured and unstructured outliers and missing data.
Recent advances in Convolutional Neural Network (CNN) based approaches have been able to generate... more Recent advances in Convolutional Neural Network (CNN) based approaches have been able to generate convincing talking heads. Personalization of such talking heads requires training of the model with a large number of examples of the target person. This is also time consuming. In this paper, we propose a meta-learning based few-shot approach for generating personalized 2D talking heads where the lip animation is driven by a given audio. The idea is that the model is meta-trained with a dataset consisting of a large variety of subjects’ ethnicity and vocabulary. We show that our meta-trained model is then capable of generating realistic animation for previously unseen face and unseen audio when finetuned with only a few-shot examples for a very short time (180 seconds). Considering the fact that facial expressions driven by audio are mainly expressed through motion around lips, we restrict ourselves to animating lip only. We have done the experiments on two publicly available datasets:...
2015 Fifth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)
IEEE Transactions on Multimedia
The Visual Computer, 2016
Multimedia Tools and Applications, 2015
Proceedings of the Eighth Indian Conference on Computer Vision Graphics and Image Processing, Dec 16, 2012
ABSTRACT Imposing expressions on expression-neutral human face images is an interesting applicati... more ABSTRACT Imposing expressions on expression-neutral human face images is an interesting application of human-computer-interaction, animation, entertainment and other such fields. The objective of this paper is to impose one of the six prototypic emotional expressions i.e., Joy, Surprise, Disgust, Fear, Anger and Sorrow to a given expression-neutral face image. For this, we first establish individual models for each of the six prototypic expressions. This model is independent of the shape and texture i.e., identity of the subjects in the training video sequences. Given an intensity of a particular expression, we find the changes in the shape and texture due to a particular expression from the derived models. These changes are added to the automatically annotated test image on which the expression is to be imposed. The major contributions of the paper are: (1) Developing a method for finding facial structure specific changes of a subject for imposing a particular expression and (2) Establishing a nonlinear relationship between the expression intensity and the corresponding facial changes. The experimental results show that the proposed method is better compared to another related method. The proposal is also good at preserving the identity of the subject while imposing a given expression on the expression-neutral face image.
Proceedings of the 2014 Indian Conference on Computer Vision Graphics and Image Processing - ICVGIP '14, 2014
2013 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition (FG), 2013
ABSTRACT In real life facial expressions show mixture of emotions. This paper proposes a novel ex... more ABSTRACT In real life facial expressions show mixture of emotions. This paper proposes a novel expression descriptor based expression map that can efficiently represent pure, mixture and transition of facial expressions. The expression descriptor is the integration of optic flow and image gradient values and the descriptor value is accumulated in temporal scale. The expression map is realized using self-organizing map. We develop an objective scheme to find the percentage of different prototypical pure emotions (e.g., happiness, surprise, disgust etc.) that mix up to generate a real facial expression. Experimental results show that the expression map can be used as an effective classifier for facial expressions.
Proceedings of the Eighth Indian Conference on Computer Vision, Graphics and Image Processing - ICVGIP '12, 2012
ABSTRACT A novel graph theoretic approach is proposed for recognizing each of the six basic proto... more ABSTRACT A novel graph theoretic approach is proposed for recognizing each of the six basic prototypic human emotional facial expressions from a video sequence. The feature used, called a Shape-Motion-Descriptor (SMD), is based on orientation-quantized gradient-weighted optical flow in a hierarchical manner. The basis of the SMD is learned using a codebook learning technique. Inter-relations between SMDs are represented through a graph. A novel definition of de-centrality measure of graph connectivity is devised to make the initially large codebook, a compact one. Each video sequence is represented by an expression descriptor. The efficiency of the proposed expression descriptor is tested using different classifiers and the results are compared with the state-of-the-art methods in literature. Our result is at par or better than competing methods.
Proceedings of the Eighth Indian Conference on Computer Vision, Graphics and Image Processing - ICVGIP '12, 2012
ABSTRACT Imposing expressions on expression-neutral human face images is an interesting applicati... more ABSTRACT Imposing expressions on expression-neutral human face images is an interesting application of human-computer-interaction, animation, entertainment and other such fields. The objective of this paper is to impose one of the six prototypic emotional expressions i.e., Joy, Surprise, Disgust, Fear, Anger and Sorrow to a given expression-neutral face image. For this, we first establish individual models for each of the six prototypic expressions. This model is independent of the shape and texture i.e., identity of the subjects in the training video sequences. Given an intensity of a particular expression, we find the changes in the shape and texture due to a particular expression from the derived models. These changes are added to the automatically annotated test image on which the expression is to be imposed. The major contributions of the paper are: (1) Developing a method for finding facial structure specific changes of a subject for imposing a particular expression and (2) Establishing a nonlinear relationship between the expression intensity and the corresponding facial changes. The experimental results show that the proposed method is better compared to another related method. The proposal is also good at preserving the identity of the subject while imposing a given expression on the expression-neutral face image.
Bioinformatics (Oxford, England), Jan 26, 2015
Alzheimer's disease (AD) is a dementia that gets worse with time resulting in loss of memory ... more Alzheimer's disease (AD) is a dementia that gets worse with time resulting in loss of memory and cognitive functions. The life expectancy of AD patients following diagnosis is approximately seven years. In 2006, researchers estimated that 0.40% of the world population (range 0.17% - 0.89%) were afflicted by AD, and that the prevalence rate would be tripled by 2050. Usually examination of brain tissues is required for definite diagnosis of AD. So, it is crucial to diagnose AD at an early stage via some alternative method. As the brain controls many functions via releasing signalling proteins through blood, we analyze blood plasma proteins for diagnosis of AD. Here we use a Radial Basis Function (RBF) network for feature selection called Feature Selection RBF (FSRBF) network for selection of plasma proteins that can help diagnosis of AD. We have identified a set of plasma proteins, smaller in size than previous study, with comparable prediction accuracy. We have also analyzed Mild...
2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG), 2013
Noise & Vibration Worldwide
With increasing awareness of the need for quieter operation of industrial machines, much effort h... more With increasing awareness of the need for quieter operation of industrial machines, much effort has been put into reducing the emitted noise levels in their design, development and operation stages. This study characterizes emitted noise from Jhiri machines as well as suggesting suitable control measures to be incorporated for noise abatement.
Proceedings of the 18th Conference on Embedded Networked Sensor Systems
In this paper, we present "ThermoTrak", a smartphone accessory based, real-time and acc... more In this paper, we present "ThermoTrak", a smartphone accessory based, real-time and accurate temperature measurement mechanism, which can be used to screen for fever, which is a manifestation of infectious diseases including the symptoms caused by SARS-CoV-2. Our system accurately identifies face and forehead region from a safe distance of one meter, calculates accurate temperature of forehead with accuracy of ±0.5° C on a linear scale. An AI based algorithm is employed for the purpose of accurately detecting the Region of interest (ROI) (Face & point near center of Forehead) and calculate the absolute temperature within 300 milliseconds.
The paper propose a cognitive inspired change detection method for the detection and localization... more The paper propose a cognitive inspired change detection method for the detection and localization of shape variations on point clouds. A well defined pipeline is introduced by proposing a coarse to fine approach: i) shape segmentation, ii) fine segment registration using attention blocks. Shape segmentation is obtained using covariance based method and fine segment registration is carried out using gravitational registration algorithm. In particular the introduction of this partition-based approach using visual attention mechanism improves the speed of deformation detection and localization. Some results are shown on synthetic data of house and aircraft models. Experimental results shows that this simple yet effective approach designed with an eye to scalability can detect and localize the deformation in a faster manner. A real world car use case is also presented with some preliminary promising results useful for auditing and insurance claim tasks.
2017 IEEE International Conference on Image Processing (ICIP), Sep 1, 2017
In this paper, a shape-constrained iterative algorithm is proposed to register a rigid template p... more In this paper, a shape-constrained iterative algorithm is proposed to register a rigid template point-cloud to a given reference point-cloud. The algorithm embeds a shape-based similarity constraint into the principle of gravitation. The shape-constrained gravitation, as induced by the reference, controls the movement of the template such that at each iteration, the template better aligns with the reference in terms of shape. This constraint enables the alignment in difficult conditions indtroduced by change (presence of outliers and/or missing parts), translation, rotation and scaling. We discuss efficient implementation techniques with least manual intervention. The registration is shown to be useful for change detection in the 3D point-cloud. The algorithm is compared with three state-of-the-art registration approaches. The experiments are done on both synthetic and real-world data. The proposed algorithm is shown to perform better in the presence of big rotation, structured and unstructured outliers and missing data.
Recent advances in Convolutional Neural Network (CNN) based approaches have been able to generate... more Recent advances in Convolutional Neural Network (CNN) based approaches have been able to generate convincing talking heads. Personalization of such talking heads requires training of the model with a large number of examples of the target person. This is also time consuming. In this paper, we propose a meta-learning based few-shot approach for generating personalized 2D talking heads where the lip animation is driven by a given audio. The idea is that the model is meta-trained with a dataset consisting of a large variety of subjects’ ethnicity and vocabulary. We show that our meta-trained model is then capable of generating realistic animation for previously unseen face and unseen audio when finetuned with only a few-shot examples for a very short time (180 seconds). Considering the fact that facial expressions driven by audio are mainly expressed through motion around lips, we restrict ourselves to animating lip only. We have done the experiments on two publicly available datasets:...
2015 Fifth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG)
IEEE Transactions on Multimedia
The Visual Computer, 2016
Multimedia Tools and Applications, 2015
Proceedings of the Eighth Indian Conference on Computer Vision Graphics and Image Processing, Dec 16, 2012
ABSTRACT Imposing expressions on expression-neutral human face images is an interesting applicati... more ABSTRACT Imposing expressions on expression-neutral human face images is an interesting application of human-computer-interaction, animation, entertainment and other such fields. The objective of this paper is to impose one of the six prototypic emotional expressions i.e., Joy, Surprise, Disgust, Fear, Anger and Sorrow to a given expression-neutral face image. For this, we first establish individual models for each of the six prototypic expressions. This model is independent of the shape and texture i.e., identity of the subjects in the training video sequences. Given an intensity of a particular expression, we find the changes in the shape and texture due to a particular expression from the derived models. These changes are added to the automatically annotated test image on which the expression is to be imposed. The major contributions of the paper are: (1) Developing a method for finding facial structure specific changes of a subject for imposing a particular expression and (2) Establishing a nonlinear relationship between the expression intensity and the corresponding facial changes. The experimental results show that the proposed method is better compared to another related method. The proposal is also good at preserving the identity of the subject while imposing a given expression on the expression-neutral face image.
Proceedings of the 2014 Indian Conference on Computer Vision Graphics and Image Processing - ICVGIP '14, 2014
2013 10th IEEE International Conference and Workshops on Automatic Face and Gesture Recognition (FG), 2013
ABSTRACT In real life facial expressions show mixture of emotions. This paper proposes a novel ex... more ABSTRACT In real life facial expressions show mixture of emotions. This paper proposes a novel expression descriptor based expression map that can efficiently represent pure, mixture and transition of facial expressions. The expression descriptor is the integration of optic flow and image gradient values and the descriptor value is accumulated in temporal scale. The expression map is realized using self-organizing map. We develop an objective scheme to find the percentage of different prototypical pure emotions (e.g., happiness, surprise, disgust etc.) that mix up to generate a real facial expression. Experimental results show that the expression map can be used as an effective classifier for facial expressions.
Proceedings of the Eighth Indian Conference on Computer Vision, Graphics and Image Processing - ICVGIP '12, 2012
ABSTRACT A novel graph theoretic approach is proposed for recognizing each of the six basic proto... more ABSTRACT A novel graph theoretic approach is proposed for recognizing each of the six basic prototypic human emotional facial expressions from a video sequence. The feature used, called a Shape-Motion-Descriptor (SMD), is based on orientation-quantized gradient-weighted optical flow in a hierarchical manner. The basis of the SMD is learned using a codebook learning technique. Inter-relations between SMDs are represented through a graph. A novel definition of de-centrality measure of graph connectivity is devised to make the initially large codebook, a compact one. Each video sequence is represented by an expression descriptor. The efficiency of the proposed expression descriptor is tested using different classifiers and the results are compared with the state-of-the-art methods in literature. Our result is at par or better than competing methods.
Proceedings of the Eighth Indian Conference on Computer Vision, Graphics and Image Processing - ICVGIP '12, 2012
ABSTRACT Imposing expressions on expression-neutral human face images is an interesting applicati... more ABSTRACT Imposing expressions on expression-neutral human face images is an interesting application of human-computer-interaction, animation, entertainment and other such fields. The objective of this paper is to impose one of the six prototypic emotional expressions i.e., Joy, Surprise, Disgust, Fear, Anger and Sorrow to a given expression-neutral face image. For this, we first establish individual models for each of the six prototypic expressions. This model is independent of the shape and texture i.e., identity of the subjects in the training video sequences. Given an intensity of a particular expression, we find the changes in the shape and texture due to a particular expression from the derived models. These changes are added to the automatically annotated test image on which the expression is to be imposed. The major contributions of the paper are: (1) Developing a method for finding facial structure specific changes of a subject for imposing a particular expression and (2) Establishing a nonlinear relationship between the expression intensity and the corresponding facial changes. The experimental results show that the proposed method is better compared to another related method. The proposal is also good at preserving the identity of the subject while imposing a given expression on the expression-neutral face image.
Bioinformatics (Oxford, England), Jan 26, 2015
Alzheimer's disease (AD) is a dementia that gets worse with time resulting in loss of memory ... more Alzheimer's disease (AD) is a dementia that gets worse with time resulting in loss of memory and cognitive functions. The life expectancy of AD patients following diagnosis is approximately seven years. In 2006, researchers estimated that 0.40% of the world population (range 0.17% - 0.89%) were afflicted by AD, and that the prevalence rate would be tripled by 2050. Usually examination of brain tissues is required for definite diagnosis of AD. So, it is crucial to diagnose AD at an early stage via some alternative method. As the brain controls many functions via releasing signalling proteins through blood, we analyze blood plasma proteins for diagnosis of AD. Here we use a Radial Basis Function (RBF) network for feature selection called Feature Selection RBF (FSRBF) network for selection of plasma proteins that can help diagnosis of AD. We have identified a set of plasma proteins, smaller in size than previous study, with comparable prediction accuracy. We have also analyzed Mild...
2013 Fourth National Conference on Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG), 2013