Qualification: 
M.Tech, B-Tech

Rimjhim Padam Singh currently serves as an Assistant Professor in the Department of Computer Science, School of Engineering, Amrita Vishwa Vidyapeetham, Coimbatore. She has submitted her Ph.D. thesis on “A Light-weight Sample Consensus Based Approach for Efficient Change Detection in Video Sequence Images” to Visvesvaraya National Institute of Technology, Nagpur. She received her B. Tech. and M.Tech. Degree from RTM Nagpur University in 2013 and 2015 respectively. Her research areas include Image and Video Processing, Motion Analysis and Text Mining.

Publications

Publication Type: Conference Paper

Year of Publication Title

2020

J. Madarkar, Sharma, P., and Rimjhim Padam Singh, “Improved Performance and Execution Time of Face Recognition Using MRSRC”, in Soft Computing for Problem Solving, Singapore, 2020.[Abstract]


Madarkar, JitendraSharma, PoonamSingh, RimjhimFace recognition accuracy is vulnerable to environmental noise, low-resolution images, and other variations such as illumination, pose, and expression. The accuracy of the face recognition mostly relying on the features of training samples and testing samples. Recently, sparse representation based classification (SRC) has shown state-of-the-art results in face recognition and developed several extended versions of SRC methods to improve the performance. The time complexity of the SRC is depended on the size of the dictionary. In this paper, a new fusion approach MRSRC (Multi-resolution sparse representation based classification) is developed by incorporating the wavelet compressed features into the dictionary. MRSRC has shown better performance than an existing algorithm and also reduces the time complexity. The experimentation is carried out on benchmarking databases such as LFW and ORL.

More »»

2019

Rimjhim Padam Singh and Sharma, P., “Improving Change Detection Using Centre-Symmetric Local Binary Patterns”, in Pattern Recognition and Machine Intelligence, Cham, 2019.[Abstract]


Efficient change detection in real-time applications is a major research goal in computer vision. Several researchers have put efforts in this direction and have achieved notable performances in varied challenging situations. But handling all the challenges posed in real-time environments with a single change detection method is almost impracticable. On the other hand, ensemble based background modelling techniques have obtained improved results but they also suffer from trade-off between efficiency and hardware or time requirements, thereby hindering their real-time applicability. This paper proposes an effective hybrid change detection algorithm, light and simple enough to have an effective real-time applicability. The proposed hybrid change detection algorithm employs per-channel RGB colour features with centre-symmetric local binary patterns for pixel-modelling and feeds it to a sample-consensus classification technique for foreground segmentation. Finally, performance of the proposed technique has been tested on widely accepted change detection dataset namely, 2014 Change detection dataset (2014 CDnet dataset).

More »»

Publication Type: Journal Article

Year of Publication Title

2020

Rimjhim Padam Singh and Sharma, P., “Instance-vote-based Motion Detection using Spatially Extended Hybrid Feature Space”, Accepted at The Visual Computer Journal, 2020.[Abstract]


Motion recognition, a trivial step employed in several video-based applications, is still a challenging task in real-world complex 1 scenarios containing dynamic noise, varying backgrounds, shadows, improper illuminations, camouflages, etc. Numerous pixel-based change detection techniques employing varied combinations of different feature spaces have been proposed to 2 efficiently overcome many real-world challenges. But ideally, handling all the possible real-world challenges simultaneously is yet to be achieved. Hence, this paper proposes a memory-efficient unique combination of multi-colour feature space with a light-weight intensity-based texture descriptor. The proposed spatially enlarged extended centre-symmetric local binary pattern is combined with YCbCR and RGB colour features for robust pixel representation. The proposed feature space is fed to an extended instance-vote technique for pixel classification. The random and time-subsampled update is employed conditionally for model update, followed by a feedback network that continuously optimizes the local threshold and learning 3 rate parameters of the proposed model. The proposed feature space and model have been evaluated on whole 2014 Change Detection dataset, the largest known dataset. The outperforming performance and memory analysis strengthens its acceptability 4 for real-time applications.

More »»

2019

Rimjhim Padam Singh, Sharma, P., and Madarkar, J., “Compute-Extensive Background Subtraction for Efficient Ghost Suppression”, IEEE Access, vol. 7, pp. 130180-130196, 2019.[Abstract]


Efficient background modelling has always been an active area of research due to its immense importance as a preliminary step in various machine-vision applications. Several techniques have been proposed to date that strive to achieve higher accuracy without compromising on computational and hardware demands. One of such techniques, Visual Background Extractor (Vibe), has set benchmarks due to its fewer memory requirements and good results. However, it suffers from high false positives due to its slower, selective and random update policy. This paper proposes a novel sample-consensus pixel-based technique for efficient foreground segmentation complemented with faster ghost suppression. This is achieved by employing segmentation masks exploiting both static and dynamic properties of pixels depicting likeliness towards absorption into the foregrounds. Dynamic characteristics of the proposed approach handle `object present in the first frame' problem while static characteristics handle improper illumination and shadows in videos in lesser time. It aims not only at suppressing ghosts in the foreground mask but also allows their absorption by updating the background model with such regions. It also proposes a unique spatio-temporal model initialization technique for handling continuous noise. The proposed approach proved to produce outstanding results when compared with 9 traditional and 13 state-of-the-art algorithms.

More »»

2019

Rimjhim Padam Singh and Sharma, P., “A Light-Weight Change Detection Method Using YCbCr-Based Texture Consensus Model”, International Journal of Pattern Recognition and Artificial Intelligence, p. 2050023, 2019.[Abstract]


Background subtraction is a prerequisite and often the very first step employed in several high-level and real-time computer vision applications. Several parametric and non-parametric change detection algorithms employing multiple feature spaces have been proposed to date but none has proven to be robust against all challenges that can possibly be posed in a complex real-time environment. Amongst the varied challenges posed, illumination variations, shadows, dynamic backgrounds, camouflaged and bootstrapping artifacts are some of the well-known problems. This paper presents a light-weight hybrid change detection algorithm that integrates a novel combination of RGB color space and conditional YCbCr-based XCS-LBP texture descriptors (YXCS-LBP) into a modified pixel-based background model. The conditional employment of light-weight YXCS-LBP texture features with the modified Visual background extractor (ViBe) aiming at reduction in false positives, produces outperforming results without incurring much memory and computational cost. The random and time-subsampled update strategy employed with the proposed classification procedure ensures the efficient suppression of shadows and bootstrapping artifacts along with the complete retention of long-term static objects in the foreground masks. Comprehensive performance analysis of the proposed technique on publicly available Change Detection dataset (2014 CDnet dataset) demonstrates the superiority of the proposed technique over different state-of-the-art-methods against varied challenges.

More »»

2018

Rimjhim Padam Singh and Chandak, M. B., “A Survey on Various Strategies for Classification and Novel Class Detection of Data Streams”, International Journal Of Computer Science And Applications , vol. 8, 2018.[Abstract]


Data Stream: A continuous stream of raw data has to be converted into some intelligible form to extract meaningful information from it. Only then the data can be put to use. Handling real time data streams in data mining isn't easy and poses various challenges to researchers. It poses four main challenges namely, Infinite length, Concept-drift, Concept-evolution and Feature evolution. Various researchers have proposed various techniques for overcoming these difficulties. But major work has been done to handle infinite length and conceptdrift in data streams only. The other two problems have not been tackled that efficiently. In this paper, we make an effort to list and summarize various strategies that have been proposed to overcome the above stated problems for efficiently making use of stream data.

More »»

2015

Rimjhim Padam Singh, “Classification and Novel Class Detection in Data Streams Using Strings”, Open Access Library Journal 2, 2015.[Abstract]


Data streams are continuous and always keep evolving in nature. Because of these reasons it becomes difficult to handle such data with simple and static strategies. Data stream poses four main challenges to researchers. These are infinite length, concept-evolution, concept-drift and feature evolution. Infinite-length is because of the amount of data having no bounds. Concept-drift is due to slow changes in the concept of stream. Concept-evolution occurs due to presence of unknown classes in data. Feature-evolution is because of new features continuously keeping appearing in the stream and older ones start disappearing. For performing any analysis on such data we first need to convert it into some knowledgeable form and also need to handle the above mentioned challenges. Various strategies have been proposed to tackle these difficulties. But most of them focus on handling the problem of infinite-length and concept-drift. In this paper, we make efforts to propose a string based strategy to handle infinite-length, concept-evolution and concept-drift.

More »»

Publication Type: Conference Proceedings

Year of Publication Title

2018

Rimjhim Padam Singh, Sharma, P., and Madarkar, J., “Motion Detection Using a Hybrid Texture-Based Approach”, In Soft Computing for Problem Solving. Springer, Singapore, pp. 609-620, 2018.[Abstract]


Motion analysis plays an important role in various real-time applications like object detection, human–computer interaction, surveillance systems, human detection and tracking, event monitoring, etc. Background subtraction that aims at separating the motion regions from the static portions lays the foundation of all such applications. Most of the background subtraction techniques developed to date explore colour features of pixels, either individually or in a spatio-temporal manner. Many other techniques exploit texture characteristics of pixels, while a few have been developed that employ a combination of both texture and colour characteristics for extracting motion-related information from frames. But most of the efficient background modelling techniques demand extensive use of hardware and computation. In this paper, we propose a hybrid sample consensus-based foreground segmentation technique that fuses similarity-based binary patterns of pixels with YCbCr colour space. The core of a pixel-based technique has been reconstructed to obtain drastically refined results.

More »»

Poster Presentations

Rimjhim Padam Singh. ”Motion Detection using a Hybrid-Texture Based Approach” at Research Scholar Day conducted by VNIT, Nagpur

Package / Software

Rimjhim Singh, Poonam Sharma, ”Rbgs: An R package on reading and background subtration in videos”, on Comprehensive R Archive Network. (https://CRAN.R-project.org/package=Rbgs).