arXiv subjects

Pravin Kumar

Publications and source records attributed to Pravin Kumar.

11 recordsLinked to original sources

arXiv2026

Smart Railway Obstruction Detection System using IoT and Computer Vision

Railway track intrusions pose a critical safety challenge for Indian Railways, encompassing wildlife incursions and deliberate malicious obstructions. The December 2025 collision in Assam, in which seven elephants were killed by the Rajdhani Express, underscores the urgency of effective real-time detection. Existing solutions such as the optical fiber-based Gajraj system suffer from prohibitive costs (\$1000/km) and high false alarm rates, limiting deployment to only 20 of India's 101 elephant corridors. This paper proposes NETRA, a cost-effective, internet-independent intrusion detection system deployed on Raspberry Pi Zero W and Raspberry Pi 4 edge platforms. NETRA employs probabilistic sensor fusion integrating a PIR motion sensor and an HC-SR04 ultrasonic distance sensor with a tunable threshold (tau_c = 0.65), enabling event-driven camera activation that reduces unnecessary visual processing by 52%. Upon confirmed intrusion, edge-AI classification using MobileNet-SSD (Pi Zero) or YOLOv5 ONNX (Pi 4) identifies threats including humans, large animals, and track obstructions. Confirmed threats are transmitted via LoRa (868 MHz) to alert the locomotive driver within 2.4 seconds end-to-end. Experimental evaluation across 113 motion events demonstrated 95% detection accuracy with zero false alarms through probabilistic fusion, compared to 85% for binary methods. Raspberry Pi 4 with YOLOv5 achieved 83.5% elephant F1-score, a 5.6x improvement over Pi Zero's heuristic approach (14.8%). LoRa communication achieved 100% packet delivery across 1-2 km in field trials. NETRA reduces deployment cost by 75% (\$247/km vs \$1000/km for Gajraj) while providing unified detection of both wildlife and obstruction threats.

cs.CV↗

arXiv2026

Privacy-Preserving Sensor-Based Human Activity Recognition for Low-Resource Healthcare Using Classical Machine Learning

Limited access to medical infrastructure forces elderly and vulnerable patients to rely on home-based care, often leading to neglect and poor adherence to therapeutic exercises such as yoga or physiotherapy. To address this gap, we propose a low-cost and automated human activity recognition (HAR) framework based on wearable inertial sensors and machine learning. Activity data, including walking, walking upstairs, walking downstairs, sitting, standing, and lying, were collected using accelerometer and gyroscope measurements. Four classical classifiers, Logistic Regression, Random Forest, Support Vector Machine (SVM), and k-Nearest Neighbors (k-NN), were evaluated and compared with the proposed Support Tensor Machine (STM). Experimental results show that SVM achieved an accuracy of 93.33 percent, while Logistic Regression, Random Forest, and k-NN achieved 91.11 percent. In contrast, STM significantly outperformed these models, achieving a test accuracy of 96.67 percent and the highest cross-validation accuracy of 98.50 percent. Unlike conventional methods, STM leverages tensor representations to preserve spatio-temporal motion dynamics, resulting in robust classification across diverse activities. The proposed framework demonstrates strong potential for remote healthcare, elderly assistance, child activity monitoring, yoga feedback, and smart home wellness, offering a scalable solution for low-resource and rural healthcare settings.

cs.LG↗

arXiv2025

Bharat Scene Text: A Novel Comprehensive Dataset and Benchmark for Indian Language Scene Text Understanding

Reading scene text, that is, text appearing in images, has numerous application areas, including assistive technology, search, and e-commerce. Although scene text recognition in English has advanced significantly and is often considered nearly a solved problem, Indian language scene text recognition remains an open challenge. This is due to script diversity, non-standard fonts, and varying writing styles, and, more importantly, the lack of high-quality datasets and open-source models. To address these gaps, we introduce the Bharat Scene Text Dataset (BSTD) - a large-scale and comprehensive benchmark for studying Indian Language Scene Text Recognition. It comprises more than 100K words that span 11 Indian languages and English, sourced from over 6,500 scene images captured across various linguistic regions of India. The dataset is meticulously annotated and supports multiple scene text tasks, including: (i) Scene Text Detection, (ii) Script Identification, (iii) Cropped Word Recognition, and (iv) End-to-End Scene Text Recognition. We evaluated state-of-the-art models originally developed for English by adapting (fine-tuning) them for Indian languages. Our results highlight the challenges and opportunities in Indian language scene text recognition. We believe that this dataset represents a significant step toward advancing research in this domain. All our models and data are open source.

cs.CV↗

arXiv2025

Congruence subgroups of small Artin and Coxeter groups

Small Coxeter groups are exactly those for which the Tits representation takes integral values, which makes the study of their congruence subgroups significant. In \cite{MR0938643}, Squier introduced a matrix representation of an Artin group defined over the ring $\mathbb Z[s^{\pm}, t^{\pm}]$ of Laurent polynomials in two variables. This representation simultaneously generalises the Tits representation of the associated Coxeter groups and the reduced Burau representation of braid groups. We define small Artin groups as those for which this representation becomes integral when evaluated at $s=1$ and $t=-1$. Consequently, the study of congruence subgroups of small Artin groups extends the classical notion of congruence subgroups of braid groups, which arise from the integral reduced Burau representation. In this paper, we examine Coxeter and Artin groups that possess the congruence subgroup property and identify several of their principal congruence subgroups at small levels.

math.GR↗

arXiv2024

Linear representations, crystallographic quotients, and twisted conjugacy of virtual Artin groups

Virtual Artin groups were recently introduced by Bellingeri, Paris, and Thiel as broad generalizations of the well-known virtual braid groups. For each Coxeter graph $\Gamma$, they defined the virtual Artin group $VA[\Gamma]$, which is generated by the corresponding Artin group $A[\Gamma]$ and the Coxeter group $W[\Gamma]$, subject to certain mixed relations inspired by the action of $W[\Gamma]$ on its root system $\Phi[\Gamma]$. There is a natural surjection $ \mathrm{VA}[\Gamma] \rightarrow W[\Gamma]$, with the kernel $PVA[\Gamma]$ representing the pure virtual Artin group. In this paper, we explore linear representations, crystallographic quotients, and twisted conjugacy of virtual Artin groups. Inspired from the work of Cohen, Wales, and Krammer, we construct a linear representation of the virtual Artin group $VA[\Gamma]$. As a consequence of this representation, we deduce that if $W[\Gamma]$ is a spherical Coxeter group, then $VA[\Gamma]/PVA[\Gamma]'$ is a crystallographic group of dimension $ |\Phi[\Gamma]|$ with the holonomy group $W[\Gamma]$. We also classify the torsion elements in $VA[\Gamma]/PVA[\Gamma]'$ and determine precisely when two elements are conjugate in this group. Further, we investigate twisted conjugacy, and prove that each right-angled virtual Artin group admit the $R_\infty$-property.

math.GR↗

arXiv2024

Orderability of big mapping class groups

We give an alternate proof of the left-orderability of the mapping class group of a connected oriented infinite-type surface with a non-empty boundary. Our main strategy involves the inductive construction of a countable stable Alexander system for the surface using a carefully chosen exhaustion by finite-type subsurfaces. In fact, we prove that a generalised ideal arc system for the surface also induces a left-ordering on the big mapping class group. We then prove that two generalised ideal arc systems determine the same left-ordering if and only if they are loosely isotopic. Finally, we prove that the topology on the big mapping class group is the same as the order topology induced by a left-ordering corresponding to an inductively constructed ideal arc system.

math.GT↗

arXiv2023

Brunnian planar braids and simplicial groups

Twin groups are planar analogues of Artin braid groups and play a crucial role in the Alexander-Markov correspondence for the isotopy classes of immersed circles on the 2-sphere without triple and higher intersections. These groups admit diagrammatic representations, leading to maps obtained by the addition and deletion of strands. This paper explores Brunnian twin groups, which are subgroups of twin groups composed of twins that become trivial when any of their strands are deleted. We establish that Brunnian twin groups consisting of more than two strands are free groups. Furthermore, we provide a necessary and sufficient condition for a Brunnian doodle on the 2-sphere to be the closure of a Brunnian twin. Additionally, we delve into two generalizations of Brunnian twins, namely, $k$-decomposable twins and Cohen twins, and prove some structural results about these groups. We also investigate a simplicial structure on pure twin groups that admits a simplicial homomorphism from Milnor's construction of the simplicial 2-sphere. This gives a possibility to provide a combinatorial description of homotopy groups of the 2-sphere in terms of pure twins.

math.GR↗

arXiv2023

Twisted Rokhlin property for mapping class groups

In this paper, generalising the idea of the Rokhlin property, we explore the concept of the twisted Rokhlin property of topological groups. A topological group is said to exhibit the twisted Rokhlin property if, for each automorphism $\phi$ of the group, there exists a $\phi$-twisted conjugacy class that is dense in the group. We provide a complete classification of connected orientable infinite-type surfaces without boundaries whose mapping class groups possess the twisted Rokhlin property. Additionally, we prove that the mapping class groups of the remaining surfaces do not admit any dense $\phi$-twisted conjugacy class for any automorphism $\phi$. This supplements the recent work of Lanier and Vlamis on the Rokhlin property of big mapping class groups. We also prove that the mapping class group of each connected orientable infinite-type surface without boundary possesses the $R_\infty$-property.

math.GT↗

arXiv2023

Commutator subgroups and crystallographic quotients of virtual extensions of symmetric groups

The virtual braid group $VB_n$, the virtual twin group $VT_n$ and the virtual triplet group $VL_n$ are extensions of the symmetric group $S_n$, which are motivated by the Alexander-Markov correspondence for virtual knot theories. The kernels of natural epimorphisms of these groups onto the symmetric group $S_n$ are the pure virtual braid group $VP_n$, the pure virtual twin group $PVT_n$ and the pure virtual triplet group $PVL_n$, respectively. In this paper, we investigate commutator subgroups, pure subgroups and crystallographic quotients of these groups. We derive explicit finite presentations of the pure virtual triplet group $PVL_n$, the commutator subgroup $VT_n^{'}$ of $VT_n$ and the commutator subgroup $VL_n^{'}$ of $VL_n$. Our results complete the understanding of these groups, except that of $VB_n^{'}$, for which the existence of a finite presentations is not known for $n \ge 4$. We also prove that $VL_n/PVL_n^{'}$ is a crystallographic group and give an explicit construction of infinitely many torsion elements in it.

math.GR↗

arXiv2022

Congruence subgroups and crystallographic quotients of small Coxeter groups

Small Coxeter groups are precisely the ones for which the Tits representation is integral, which makes the study of their congruence subgroups relevant. The symmetric group $S_n$ has three natural extensions, namely, the braid group $B_n$, the twin group $T_n$ and the triplet group $L_n$. The latter two groups are small Coxeter groups, and play the role of braid groups under the Alexander-Markov correspondence for appropriate knot theories, with their pure subgroups admitting suitable hyperplane arrangements as Eilenberg-MacLane spaces. In this paper, we prove that the congruence subgroup property fails for infinite small Coxeter groups which are not virtually abelian. As an application, we deduce that the congruence subgroup property fails for both $T_n$ and $L_n$ when $n \ge 4$. We also determine subquotients of principal congruence subgroups of $T_n$, and identify the pure twin group $PT_n$ and the pure triplet group $PL_n$ with suitable principal congruence subgroups. Further, we investigate crystallographic quotients of these two families of small Coxeter groups, and prove that $T_n /PT_n^{'}$, $T_n/T_n^{''}$ and $L_n /PL_n^{'}$ are crystallographic groups. We also determine crystallographic dimensions of these groups and identify the holonomy representation of $T_n/T_n^{''}$.

math.GR↗

arXiv2019

Unusual charge exchange by swift heavy ions at solid surfaces

We have employed x-ray spectroscopy to probe the charge changing process only in the bulk of the foil when swift heavy ions pass through it. In contrast, the electromagnetic methods take into account integral effect of the charge changing process in the bulk as well as the charge exchange phenomenon at the surface of the foil. Thus, the difference between the mean charge states so measured from the two methods disentangles the charge exchange phenomenon at the surface from the charge changing process in the bulk and, provides opportunities to refine the understanding of ion-surface interactions. Very surprisingly, up to tens of electrons per event participate in the charge exchange phenomenon during swift heavy ion-surface interactions. This finding has been validated with a series of experiments using several ions (z = 22-35) in the energy range of 1.5-3.0 MeV/u and also verified theoretically with Fermi-gas model. Interestingly, such unusual charge exchange phenomenon could play significant role in x-ray emission of many astrophysical environments, infrared emission bands from range of environments in galaxies, accelerator physics, ion energy losses in solids, heavy ion cancer treatments, inner shell ionization by heavy ions, and surface modifications in nano scale.

physics.atom-ph↗