arXiv subjects

Arindam Das

Publications and source records attributed to Arindam Das.

At least 73 records · Page 4Linked to original sources

arXiv2022

Spatio-Contextual Deep Network Based Multimodal Pedestrian Detection For Autonomous Driving

Pedestrian Detection is the most critical module of an Autonomous Driving system. Although a camera is commonly used for this purpose, its quality degrades severely in low-light night time driving scenarios. On the other hand, the quality of a thermal camera image remains unaffected in similar conditions. This paper proposes an end-to-end multimodal fusion model for pedestrian detection using RGB and thermal images. Its novel spatio-contextual deep network architecture is capable of exploiting the multimodal input efficiently. It consists of two distinct deformable ResNeXt-50 encoders for feature extraction from the two modalities. Fusion of these two encoded features takes place inside a multimodal feature embedding module (MuFEm) consisting of several groups of a pair of Graph Attention Network and a feature fusion unit. The output of the last feature fusion unit of MuFEm is subsequently passed to two CRFs for their spatial refinement. Further enhancement of the features is achieved by applying channel-wise attention and extraction of contextual information with the help of four RNNs traversing in four different directions. Finally, these feature maps are used by a single-stage decoder to generate the bounding box of each pedestrian and the score map. We have performed extensive experiments of the proposed framework on three publicly available multimodal pedestrian detection benchmark datasets, namely KAIST, CVC-14, and UTokyo. The results on each of them improved the respective state-of-the-art performance. A short video giving an overview of this work along with its qualitative results can be seen at https://youtu.be/FDJdSifuuCs. Our source code will be released upon publication of the paper.

cs.CV↗

arXiv2021

Bounds on the triplet fermions in type-III seesaw and implications for collider searches

Type-III seesaw is a simple extension of the Standard Model~(SM) with the SU$(2)_\text{L}$ triplet fermion with zero hypercharge. It can explain the origin of the tiny neutrino mass and flavor mixing. After the electroweak symmetry breaking the light neutrino mass is generated by the seesaw mechanism which further ensures the mixings between the light neutrino and heavy neutral lepton mass eigenstates. If the triplet fermions are around the electroweak scale having sizable mixings with the SM sector allowed by the correct gauge symmetry, they can be produced at the high energy colliders leaving a variety of characteristic signatures. Based on a simple and concrete realizations of the model we employ a general parametrization for the neutrino Dirac mass matrix and perform a parameter scan to identify the allowed regions satisfying the experimental constraints from the neutrino oscillation data, the electroweak precision measurements and the lepton-flavor violating processes, respectively considering the normal and inverted neutrino mass hierarchies. These parameter regions can be probed at the different collider experiments.

hep-ph↗

arXiv2020

Neutrino charge constraints from scattering to the weak gravity conjecture to neutron stars

In various extensions of the Standard Model of particle physics, and intriguingly even in the three-generation Standard Model without neutrino masses, neutrinos are allowed to have very tiny electric charges. After a review of the theoretical scenarios that allow the emergence of such charges, we discuss the existing observational limits and we derive new stringent direct upper bounds for the charges of the muon and tau neutrinos. We also point out a flavor-universal lower bound on neutrino charges which is obtained from the weak gravity conjecture, that is based on the hypothesis that gravity is the weakest force. We finally present a new flavor-universal upper bound on neutrino charges based on astrophysical observations of Magnetars.

hep-ph↗

arXiv2020

Alternative dark matter phenomenology in a general $U(1)_X$ extension of the Standard Model

The existence of the neutrino mass and flavor mixing have been experimentally verified. These phenomena strongly motivate to extend the Standard Model (SM). Amongst many possibilities, a simple and interesting extension of the SM can be investigated using a general U$(1)_X$ extension of the SM gauge group. Demanding the cancellation of the gauge and mixed gauge gravity anomalies, three right handed neutrinos are introduced in this model where the U$(1)_X$ charge assignment becomes a linear combination of U$(1)_{\rm{B-L}}$ and U$(1)_Y$ hyper-charges. After the U$(1)_X$ breaking, an additional neutral gauge boson, $Z^\prime$ is evolved and the neutrino mass is generated by the seesaw mechanism. In such a model we investigate the properties of a Dark Matter (DM) candidate which is a massive weakly interacting particle and Dirac type in nature. The stability of the DM is protected by its U$(1)_X$ charge. Using the current bounds on the search results of $Z^\prime$ at the Large Hadron Collider (LHC) and the dark matter relic abundance we find a phenomenologically viable parameter space of our scenario.

hep-ph↗

arXiv2020

Testing triplet fermions at the electron-positron and electron-proton colliders using fat jet signatures

The addition of $SU(2)_L$ triplet fermions of zero hypercharge with the Standard Model (SM) helps to explain the origin of the neutrino mass by the so-called seesaw mechanism. Such a scenario is commonly know as the type-III seesaw model. After the electroweak symmetry breaking the mixings between the light and heavy mass eigenstates of the neutral leptons are developed which play important roles in the study of the charged and neutral multiplets of the triplet fermions at the colliders. In this article we study such interactions to produce these multiplets of the triplet fermion at the electron-positron and electron-proton colliders at different center of mass energies. We focus on the heavy triplets, for example, having mass in the TeV scale so that their decay products including the SM the gauge bosons or Higgs boson can be sufficiently boosted, leading to a fat jet. Hence we probe the mixing between light-heavy mass eigenstates of the neutrinos and compare the results with the bounds obtained by the electroweak precision study.

hep-ph↗

arXiv2020

TiledSoilingNet: Tile-level Soiling Detection on Automotive Surround-view Cameras Using Coverage Metric

Automotive cameras, particularly surround-view cameras, tend to get soiled by mud, water, snow, etc. For higher levels of autonomous driving, it is necessary to have a soiling detection algorithm which will trigger an automatic cleaning system. Localized detection of soiling in an image is necessary to control the cleaning system. It is also necessary to enable partial functionality in unsoiled areas while reducing confidence in soiled areas. Although this can be solved using a semantic segmentation task, we explore a more efficient solution targeting deployment in low power embedded system. We propose a novel method to regress the area of each soiling type within a tile directly. We refer to this as coverage. The proposed approach is better than learning the dominant class in a tile as multiple soiling types occur within a tile commonly. It also has the advantage of dealing with coarse polygon annotation, which will cause the segmentation task. The proposed soiling coverage decoder is an order of magnitude faster than an equivalent segmentation decoder. We also integrated it into an object detection and semantic segmentation multi-task model using an asynchronous back-propagation algorithm. A portion of the dataset used will be released publicly as part of our WoodScape dataset to encourage further research.

cs.CV↗

arXiv2020

Looking for Minimal Inverse Seesaw scenarios at the LHC with Jet Substructure Techniques

Simple extensions of the Standard Model (SM) with additional Right Handed Neutrinos (RHNs) can elegantly explain the existence of small neutrino masses and their flavor mixings. Collider searches for sterile neutrinos are being actively pursued currently. Heavy RHNs may dominantly decay into $W^\pm l^\pm$ after being produced at the LHC. In this paper, we consider collider signatures of heavy pseudo-Dirac neutrinos in the context of inverse seesaw scenario, with a sizable mixing with the SM neutrinos under two different flavor structures, viz., Flavor Diagonal (FD) and Flavor Non-Diagonal (FND) scenarios. For the latter scenario we use a general parametrization for the model parameters by introducing an arbitrary orthogonal matrix and nonzero Dirac and Majorana phases. We then perform a parameter scan to identify allowed parameter regions which satisfy all experimental constraints. As an alternative channel to the traditional trilepton signature, we propose the opposite-sign di-lepton signature in the final state, in association with a fat jet from the hadronic decay of the boosted $W^\pm$ . We specifically consider a fat jet topology and explore the required enhancements from exploiting the characteristics of the jet substructure techniques. We perform a comprehensive collider analysis to demonstrate the effectiveness of this channel in both of the scenarios, significantly enhancing the bounds on the RHN mass and mixing angles at the 13 TeV LHC. Interestingly the FND scenario can reach up to a 5- $σ$ limit under the presence of the general parametrization at the high luminosity LHC.

hep-ph↗

arXiv2020

Radiative generation of neutrino masses in a 3-3-1 type model

A new model for tiny neutrino masses is proposed in the gauge theory of $SU(3)_C \otimes SU(3)_L \otimes U(1)_X$, where neutrino masses are generated via the quantum effect of new particles. In this model, the fermion content is taken to be minimal to realize the gauge anomaly cancellation, while the scalar sector is extended from the minimal 3-3-1 model to have an additional $SU(3)_L$ triplet field. After $SU(3)_L\otimes U(1)_X$ is broken into $SU(2)_L\otimes U(1)_Y$, the "Zee model" like diagrams are naturally induced, which contain sufficient lepton flavor violating interactions to reproduce current neutrino oscillation data. Furthermore, the remnant $Z_2$ symmetry appears after the electroweak symmetry breaking, which guarantees the stability of dark matter. It is confirmed that this model can satisfy current dark matter data. As an important prediction to test this model, productions and decays of doubly-charged scalar bosons at collider experiments are discussed in successful benchmark scenarios.

hep-ph↗

arXiv2019

Displaced heavy neutrinos from $Z'$ decays at the LHC

We study the LHC sensitivity to probe a long-lived heavy neutrino $N$ in the context of $Z'$ models. We focus on displaced vertex signatures of $N$ when pair produced via a $Z'$, decaying to leptons and jets inside the inner trackers of the LHC experiments. We explore the LHC reach with current long-lived particle search strategies for either one or two displaced vertices in association with hadronic tracks or jets. We focus on two well-motivated models, namely, the minimal $U(1)_{B-L}$ scenario and its $U(1)_{X}$ extension. We find that searches for at least one displaced vertex can cover a significant portion of the parameter space, with light-heavy neutrino mixings as low as $|V_{lN}|^2\approx 10^{-17}$, and $l=e,μ$ accessible across GeV scale heavy neutrino masses.

hep-ph↗

arXiv2019

SoildNet: Soiling Degradation Detection in Autonomous Driving

In the field of autonomous driving, camera sensors are extremely prone to soiling because they are located outside of the car and interact with environmental sources of soiling such as rain drops, snow, dust, sand, mud and so on. This can lead to either partial or complete vision degradation. Hence detecting such decay in vision is very important for safety and overall to preserve the functionality of the "autonomous" components in autonomous driving. The contribution of this work involves: 1) Designing a Deep Convolutional Neural Network (DCNN) based baseline network, 2) Exploiting several network remodelling techniques such as employing static and dynamic group convolution, channel reordering to compress the baseline architecture and make it suitable for low power embedded systems with nearly 1 TOPS, 3) Comparing various result metrics of all interim networks dedicated for soiling degradation detection at tile level of size 64 x 64 on input resolution 1280 x 768. The compressed network, is called SoildNet (Sand, snOw, raIn/dIrt, oiL, Dust/muD) that uses only 9.72% trainable parameters of the base network and reduces the model size by more than 7 times with no loss in accuracy

cs.CV↗

arXiv2019

Rejection-Cascade of Gaussians: Real-time adaptive background subtraction framework

Background-Foreground classification is a well-studied problem in computer vision. Due to the pixel-wise nature of modeling and processing in the algorithm, it is usually difficult to satisfy real-time constraints. There is a trade-off between the speed (because of model complexity) and accuracy. Inspired by the rejection cascade of Viola-Jones classifier, we decompose the Gaussian Mixture Model (GMM) into an adaptive cascade of Gaussians(CoG). We achieve a good improvement in speed without compromising the accuracy with respect to the baseline GMM model. We demonstrate a speed-up factor of 4-5x and 17 percent average improvement in accuracy over Wallflowers surveillance datasets. The CoG is then demonstrated to over the latent space representation of images of a convolutional variational autoencoder(VAE). We provide initial results over CDW-2014 dataset, which could speed up background subtraction for deep architectures.

stat.ML↗

arXiv2019

Measuring the heavy neutrino oscillations in rare W boson decays at the Large Hadron Collider

Majorana neutrinos in the seesaw model can have sizable mixings through which they can be produced at the Large Hadron Collider (LHC) and show a remarkable Lepton Number Violating (LNV) signature. In this article we study the LNV decay of the W boson via two almost degenerate heavy on-shell Majorana neutrinos $N_j$, into three charged leptons and a light neutrino. We consider the scenario where the heavy neutrino masses are within $1$ GeV $\leq M_N \leq 10$ GeV. We evaluated the possibility to measure a LNV oscillation process in such a scenario, namely, the modulation of the quantity $d Γ/d L$ for the process at the LHC where $W^{\pm} \to μ^{\pm} N \to μ^{\pm} τ^{\pm} W^{\mp *}$ $ \to μ^{\pm} τ^{\pm} e^{\mp} ν_e$. $L$ is the distance within the detector between the two vertices of the process. We found out some realistic conditions under which such a modulation could be probed at the LHC.

hep-ph↗

arXiv2019

Sensitivity of Lepton Number Violating Meson Decays in Different Experiments

We study the discovery prospect of different three body lepton number violating~(LNV) meson decays $M_{1}^{-}\to\ell_{1}^{-}\ell_{2}^{-}M_{2}^{+}$ in the framework of right handed~(RH) neutrino extended Standard Model~(SM). We consider a number of ongoing experiments, such as, NA62 and LHCb at CERN, Belle II at SuperKEK, as well as at the proposed future experiments, SHiP, MATHUSLA and FCC-ee. The RH Majorana neutrino $N$ mediating these meson decays provides a resonant enhancement of the rates, if the mass of $N$ lies in the range $(100\, \text{MeV}-6\, \text{GeV})$. We consider the effect of parent mesons velocity, as well as, the effect of finite detector size. Using the expected upper limits on the number of events for the LNV decay modes, $M_{1}^{-} \to\ell_1^{-}\ell_2^{-}π^{+}$~($M_{1}=B, B_c,D, D_{s}\,\text{and}\,K$), we analyze the sensitivity reach of the mixing angles $|V_{e N}|^{2}$, $|V_{μN}|^{2}$, $|V_{τN}|^{2}$, $|V_{e N}V_{μN}|$, $|V_{e N}V_{τN}|$ and $|V_{μN}V_{τN}|$ as a function of heavy neutrino mass $M_{N}$. We show that, inclusion of parent meson velocity can account to a large difference for active-sterile mixing, specially for $D$, $D_s$ meson decay at SHiP and $K$ meson decay at NA62. Taking into account the velocity of the $D_s$ meson, the future beam dump experiment SHiP can probe $|V_{eN}|^2 \sim 10^{-9}$. For RH neutrino mass in between 2 - 5 GeV, MATHUSLA can provide best sensitivity reach of active-sterile mixings.

hep-ph↗

arXiv2019

Probing the seesaw mechanism at the 250 GeV ILC

We consider a gauged U(1)$_{B-L}$ (Baryon-minus-Lepton number) extension of the Standard Model (SM), which is anomaly-free in the presence of three Right-Handed Neutrinos (RHNs). Associated with the U(1)$_{B-L}$ symmetry breaking the RHNs acquire their Majorana masses and then play the crucial role to generate the neutrino mass matrix by the seesaw mechanism. Towards the experimental confirmation of the seesaw mechanism, we investigate a RHN pair production through the U(1)$_{B-L}$ gauge boson ($Z^\prime$) at the 250 GeV International Linear Collider (ILC). The $Z^\prime$ gauge boson has been searched at the Large Hadron Collider (LHC) Run-2 and its production cross section is already severely constrained. The constraint will become more stringent by the future experiments with the High-Luminosity upgrade of the LHC (HL-LHC). We find a possibility that even after a null $Z^\prime$ boson search result at the HL-LHC, the 250 GeV ILC can search for the RHN pair production through the final state with same-sign dileptons plus jets, which is a `smoking-gun' signature from the Majorana nature of RHNs. In addition, some of RHNs are long-lived and leave a clean signature with a displaced vertex. Therefore, the 250 GeV ILC can operate as not only a Higgs Factory but also a RHN discovery machine to explore the origin of the Majorana neutrino mass generation, namely the seesaw mechanism.

hep-ph↗

arXiv2019

Searching for Flavored Gauge Bosons

Standard Model may allow an extended gauge sector with anomaly-free flavored gauge symmetries, such as $L_{i} - L_{j}$, $B_{i} - L_{j}$, and $B - 3L_{i}$, where $i,j=1,2,3$ are flavor indices. We investigate phenomenological implications of the new flavored gauge boson $Z^{\prime}$ in the above three classes of gauge symmetries. Focusing on the gauge boson mass above 5 GeV, we use the lepton universality test in the $Z$ and $τ/μ$ decays, LEP searches, LHC searches, neutrino trident production bound, and LHC $Z\rightarrow 4μ$ searches to put constraints on the $g^{\prime}-M_{Z^{\prime}}$ plane. When $L_1$ is involved, the LEP bounds on the $e^{-}e^{+} \rightarrow \ell^{-}\ell^{+}$ processes give the most stringent bounds, while the LHC bound becomes the strongest constraints in the large $M_{Z^{\prime}}$ region when $B_{i}$ is involved. The bound from $Z\rightarrow 4μ$ productions, which is applicable for $L_2$-involved scenarios, provides stringent bounds in the small $M_{Z^{\prime}}$ region. One exception is the $B-3L_2$ scenario, in which case only a small region is favored due to the lepton universality.

hep-ph↗

arXiv2019

Long-lived TeV-scale right-handed neutrino production at the LHC in gauged $U(1)_X$ model

A gauged $U(1)_X$ extension of the Standard Model is a simple and consistent framework to naturally incorporate three right-handed neutrinos (RHNs) for generating the observed light neutrino masses and mixing by the type-I seesaw mechanism. We examine the collider testability of the $U(1)_X$ model, both in its minimal form with the conventional charges, as well as with an alternative charge assignment, via the resonant production of the $U(1)_X$ gauge boson ($Z^\prime$) and its subsequent decay into a pair of RHNs. We first derive an updated upper limit on the new gauge coupling $g_X$ as a function of the $Z'$-boson mass from the latest LHC dilepton searches. Then we identify the maximum possible cross section for the RHN pair-production under these constraints. Finally, we investigate the possibility of having one of the RHNs long-lived, even for a TeV-scale mass. Employing the general parametrization for the light neutrino mass matrix to reproduce the observed neutrino oscillation data, we perform a parameter scan and find a simple formula for the maximum RHN lifetime as a function of the lightest neutrino mass eigenvalue ($m_{\rm lightest}$). We find that for $m_{\rm lightest}\lesssim 10^{-5}$ eV, one of the RHNs in the minimal $U(1)_X$ scenario can be long-lived with a displaced-vertex signature which can be searched for at the LHC and/or with a dedicated long-lived particle detector, such as MATHUSLA. In other words, once a long-lived RHN is observed, we can set an upper bound on the lightest neutrino mass in this model.

hep-ph↗

arXiv2019

Heavy Neutrino Search via the Higgs boson at the LHC

In the inverse see-saw model the effective neutrino Yukawa couplings can be sizable due to a large mixing angle between the light $(ν)$and heavy neutrinos $(N)$. When the right handed neutrino $(N)$ can be lighter than the Standard Model (SM) Higgs boson $(h)$. It can be produced via the on-shell decay of the Higgs, $h\to Nν$ at a significant branching fraction at the LHC. In such a process $N$ mass can be reconstructed in its dominant $N\rightarrow W \ell$ decays. We perform an analysis on this channel and its relevant backgrounds, among which the $W+$jets background is the largest. Considering the existing mixing constraints from the Higgs and electroweak precision data, the best sensitivity of the heavy neutrino search is achieved for benchmark $N$ mass at 100 and 110 GeV for upcoming high luminosity LHC runs.

hep-ph↗

arXiv2019

Constraining a general U(1)$^\prime$ inverse seesaw model from vacuum stability, dark matter and collider

We consider a class of gauged $U(1)$ extensions of the Standard Model (SM), where the light neutrino masses are generated by an inverse seesaw mechanism. In addition to the three right handed neutrinos, we add three singlet fermions and demand an extra $Z_2$ symmetry under which, the third generations of both of the neutral fermions are odd, which in turn gives us a stable dark matter candidate. We express the $U(1)$ charges of all the fermions in terms of the U(1) charges of the standard model Higgs and the new complex scalar. We study the bounds on the parameters of the model from vacuum stability, perturbative unitarity, dark matter relic density and direct detection constraints. We also obtain the collider constraints on the $Z'$ mass and the $U(1)'$ gauge coupling. Finally we compare all the bounds on the $Z'$ mass versus the $U(1)'$ gauge coupling plane.

hep-ph↗