Papers by Dr. Priyadarshi Kanungo
Automatic Lane Detection in NH5 of Odisha
Advances in intelligent systems and computing, Aug 29, 2014
The efficacy of any intelligent transportation systems depends on efficiency of the lane detectio... more The efficacy of any intelligent transportation systems depends on efficiency of the lane detection system. This paper addressed the lane detection problem during the daytime in the NH5 of Odisha, India. In NH5, the contrast between road and lane is very low because of dust and mud layers from the side of the road and at many places, the lane markings are not visible due to natural or unnatural processes of erosion. Therefore, most of the proposed lane detection algorithms that are for the foreign roads failed to correctly detect the lanes. In this paper, we proposed a lane detection model and a new thresholding approach for correct detection of lanes in the NH5 between Khandagiri and Khurdha, Odisha, India.
Mean Feature Based Age Estimation
ICICCT 2019 – System Reliability, Quality Control, Safety, Maintenance and Management, 2019
Age estimation techniques by computer vision method aim to approximate automatically within an ag... more Age estimation techniques by computer vision method aim to approximate automatically within an age band or the exact age of an individual from their face. As one of the main human facial characteristic, ageing plays a more intricate role than other elements such as one’s health condition, life cycle and extreme weather conditions, age prediction from the facial feature is a hard task. In this paper, a mean feature based age estimation is proposed to estimate the age of a person within an age band of 5 years. The proposed method outperforms the existing methods based on the percentage of accuracy within a narrow range of age band.
A YCbCr Model Based Shadow Detection and Removal Approach On Camouflaged Images
2022 OITS International Conference on Information Technology (OCIT)
Customized YOLOv3 model for face detection in crowded images
2022 6th International Conference On Computing, Communication, Control And Automation (ICCUBEA
Efficient Lane Detection Under Low-Light Conditions Using Generative Adversarial Networks
SSRN Electronic Journal
Multiple Linear Regression Based Non-Uniform Illumination Equalization for Image Thresholding
SSRN Electronic Journal
In this paper, two novel strategies have been proposed to segment the object and background in a ... more In this paper, two novel strategies have been proposed to segment the object and background in a given scene. The first one, known as Featureless (FL) approach, deals with the histogram of the original image where Parallel Genetic Algorithm (PGA) based clustering notion is used to determine the optimal threshold from the discrete nature of the histogram distribution. In this regard, we have proposed a new interconnection model for PGA. The second scheme, the Featured Based (FB) approach, is based on the proposed feature histogram distribution. A feature from the given image is extracted and the histogram corresponding to the derived feature is used to determine the optimal threshold of the original image. The proposed PGA based clustering is used to determine the optimal threshold. The performance of both the schemes is compared with that of Otsu’s and Kwon’s method and FB method is found to be the best among the three techniques.
Face Recognition from Partial Face Data
2021 International Conference in Advances in Power, Signal, and Information Technology (APSIT), 2021
During the spread of the Corona epidemic, everyone started wearing masks as protection in public ... more During the spread of the Corona epidemic, everyone started wearing masks as protection in public places. Therefore, this causes a major challenge in authentication and safety systems, such as face recognition systems in railway stations, airports, and payment systems based on facial recognition technologies. Face recognition systems are safer than touch-based biometric systems. However, the face recognition systems are ineffective in the presence of a face with a mask. Therefore, we developed an efficient algorithm using the MTCNN and VGGF model to improve the efficacy of face recognition systems in partially occluded face images. The proposed approach produced 90% accuracy in the top half of the facial images.
AimsGerminal matrix hemorrhage (GMH) is a disastrous clinical event for newborns. Neuroinflammati... more AimsGerminal matrix hemorrhage (GMH) is a disastrous clinical event for newborns. Neuroinflammation plays an important role in the development of neurological deficits after GMH. The purpose of this study is to investigate the anti-inflammatory role of secukinumab after GMH and its underlying mechanisms involving PKCβ/ERK/NF-κB signaling pathway.MethodsA total of 154 Sprague-Dawley P7 rat pups were used. GMH was induced by intraparenchymal injection of bacterial collagenase. Secukinumab was administered intranasally post-GMH. PKCβ activator PMA and p-ERK activator Ceramide C6 were administered intracerebroventricularly at 24h prior to GMH induction, respectively. Neurobehavioral tests, Western blot and immunohistochemistry were used to evaluate the efficacy of secukinumab in both short-term and long-term studies.ResultsEndogenous IL-17A, IL-17RA, PKCβ and p-ERK were increased after GMH. Secukinumab treatment improved short- and long-term neurological outcomes, reduced the expression...
IET Image Processing, 2018
The spontaneous proliferation of video data necessitates implementing hierarchical structures for... more The spontaneous proliferation of video data necessitates implementing hierarchical structures for various content management applications. Temporal video segmentation is the key towards such management. To address the problem of temporal segmentation, the current communication exploits the concept of psychological behaviour of the human visual system. Towards this goal an abrupt cut detection scheme has been proposed based on Weber's law which provides a strong spatial correlation among the neighbouring pixels. Thus, the authors provide a robust and unique solution for abrupt shot boundary detection when the frames are affected partially or fully by flashlight, fire and flicker, high motion associated with an object or camera. Further, they have devised a model for generating an automatic threshold, taking into account the statistics of the feature vector which quadrates itself with the variation in the contents of the video. The effectiveness of the proposed framework is validated by exhaustive comparison with few contemporary and recent approaches by using benchmark datasets TRECVID 2001, TRECVID 2002, TRECVID 2007 and some publicly available videos. The results obtained give credence to the remarkable improvement in the performance while preserving a good trade-off between missed hits and false hits as compared to the state-of-the-art methods.
Automatic Cut Detection Based Video Segmentation
2013 International Symposium on Computational and Business Intelligence, 2013
Due to the increasing demand of internet technology and availability of high performance computer... more Due to the increasing demand of internet technology and availability of high performance computers the demand for video surfing, transmission and retrieval is increasing day by day. The automatic annotation is a process which helps the browser or the user to retrieve the exact video as well as the exact shots or content based frames from a video by saving time and resources. The first step towards the automatic annotation is the temporal video segmentation. A boundary between two groups of video is defined as cut. A cut has been declared between two consecutive image frames if the two frames are sufficiently dissimilar. In this paper a Peak Change Ratio based cut detection has been addressed to detect the cuts in a video. The performance of the proposed method is superior in terms of "R (Recall)", and "F1" measure in comparison to the existing histogram based cut detection and Gargi's RGB color model based cut detection.
Development of DE based adaptive techniques for nonlinear system identification
2011 International Conference on Recent Trends in Information Systems, 2011
Nonlinear System Identification is generally used in control system, pattern recognition and opti... more Nonlinear System Identification is generally used in control system, pattern recognition and optimization problem. In past the Least Mean Square Algorithm (LMS), Recursive least square (RLS), Artificial Neural Network (ANN) and Genetic Algorithm (GA) have been successfully employed for nonlinear system identification. The LMS, RLS and ANN techniques are derivative based and hence are chances that the parameters may fall
… of the Conference on Soft Computing …, 2006
In this paper the problem of image segmentation is addressed using the notion of thresholding. A ... more In this paper the problem of image segmentation is addressed using the notion of thresholding. A new approach based on Genetic Algorithm (GA) is proposed for selection of threshold from the histogram of images. Specifically GA based crowding algorithm is proposed for determination of the peaks and valleys of the histogram. Experimental results are provided for histogram with bimodal feature, however, this technique can be extended to multi threshold selection for histograms with multimodal feature.
Dominant peak-means clustering for color image segmentation
2017 International Conference on Intelligent Computing and Control (I2C2), 2017
This paper presents a fully automatic color image segmentation. Colors of the pixels are used to ... more This paper presents a fully automatic color image segmentation. Colors of the pixels are used to partition the color space into clusters. Firstly the dominant peaks of the histograms of Red, Green, and Blue planes are detected using a peak detection algorithm. Then dominant peaks of these three histograms are used to define the initial number of clusters and their centroids. Finally a clustering algorithm is used to update the number of clusters and their centroids. Using these final cluster centroids each pixel is labeled to one of these clusters to segment the color image. The proposed algorithm is validated using Berkeley and Corel image data set. The result shows that our proposed method is faster and better in comparison to one of the land mark method JSEG.
In this paper, two novel strategies have been proposed to segment the object and background in a ... more In this paper, two novel strategies have been proposed to segment the object and background in a given scene. The first one, known as Featureless (FL) approach, deals with the histogram of the original image where Parallel Genetic Algorithm (PGA) based clustering notion is used to determine the optimal threshold from the discrete nature of the histogram distribution. In this regard, we have proposed a new interconnection model for PGA. The second scheme, the Featured Based (FB) approach, is based on the proposed feature histogram distribution. A feature from the given image is extracted and the histogram corresponding to the derived feature is used to determine the optimal threshold of the original image. The proposed PGA based clustering is used to determine the optimal threshold. The performance of both the schemes is compared with that of Otsu’s and Kwon’s method and FB method is found to be the best among the three techniques.
Complex & Intelligent Systems, 2021
Lane detection (LD) under different illumination conditions is a vital part of lane departure war... more Lane detection (LD) under different illumination conditions is a vital part of lane departure warning system and vehicle localization which are current trends in the future smart cities. Recently, vision-based methods are proposed to detect lane markers in different road situations including abnormal marker cases. However, an inclusive framework for driverless cars has not been introduced yet. In this work, a novel LD and tracking method is proposed for the autonomous vehicle in the IoT-based framework (IBF). The IBF consists of three modules which are vehicle board (VB), cloud module (CM), and the vehicle remote controller. The LD and tracking are carried out initially by the VB, and then, in case of any failure, the whole set of data is passed to CM to be processed and the results are sent to the VB to perform the appropriate action. If the CM detects a lane departure, then the autonomous vehicle is driven remotely and the VB would be restarted. In addition to the proposed framewo...
ELCVIA Electronic Letters on Computer Vision and Image Analysis, 2021
Selection of appropriate size of windows or subimages is the most important step for thresholding... more Selection of appropriate size of windows or subimages is the most important step for thresholding images with non-uniform lighting. In this paper, a novel criteria function is developed to partition images into different size of sub images appropriate for thresholding. After the partitioning, each subimage is segmented by Otsu’s thresholding approaches. The performance of the proposed method is validated on benchmark test images with different degree of uneven lighting. Based on the qualitative and quantitative measures, the proposed method is fully automatic, fast and efficient in comparison to many landmark approaches.
GMM Based Adaptive Thresholding for Uneven Lighting Image Binarization
Journal of Signal Processing Systems, 2021
An improved and low-complexity neural network model for curved lane detection of autonomous driving system
Soft Computing, 2021
Lane detection is considered as a key component for the autonomous vehicles of futuristic transpo... more Lane detection is considered as a key component for the autonomous vehicles of futuristic transport systems. The extraction and fitting problems of lane markers from the road images have been addressed in recent research studies. However, these are still ineffective under curved lane and color light conditions. Illumination changes and the road structure mainly affect the efficiency of lane detection which may lead to traffic accidents especially in case of a curved road. In this work, a novel method based on a low complexity but efficient functional link artificial neural network (FLANN) model is proposed to estimate the entire lane by interpolating the lane markers under different road scenarios. The road image is divided into regions and the extracted lane markers from each region are employed in the proposed trigonometric, polynomial, exponential, and Chebyshev functional expansion based FLANN models for the estimation of the lane curvature. The performance of each model is evaluated and tested on road images using three standard datasets. In terms of mean accuracy and computational time out of four FLANN models, the Chebyshev FLANN (CFLNN) outperforms other three proposed methods. The detection accuracy of CFLANN model is found to be 94.3% which is higher than the results reported by other three models.
Entropy feature and peak-means clustering based slowly moving object detection in head and shoulder video sequences
Journal of King Saud University - Computer and Information Sciences, 2021
Abstract With the increase in demand for video conferencing and IOT applications, efficient video... more Abstract With the increase in demand for video conferencing and IOT applications, efficient video coding standards are necessary. The performance of MPEG-4 coding scheme depends on the efficiency of the video object plane (VOP) generation methods. In head and shoulder video, such as news reading, video conferencing video sequences, the object has a very little movement in between two consecutive frames. Therefore, traditional segmentation methods could not extract the complete VOP efficiently. In this paper, we propose an efficient spatiotemporal segmentation method to extract the moving object for the generation of VOP in head and shoulder video sequences. First, a motion map of the object at each frame is generated based on the entropy of the temporal change of each pixel. Secondly, each frame is spatially segmented based on peak-means clustering approach. Finally, both motion map and spatial segmentation information are fused to extract the complete shape of the slow moving object. Experimental outcome depicts that the proposed method has highest detection accuracy with average intersection of union (IOU) score of 94.32% per frame and F1 measure of 97.75% per frame.
Uploads
Papers by Dr. Priyadarshi Kanungo