Search

Infrared imaging offers resilience against changing lighting conditions by capturing object temperatures. Yet, in few scenarios, its lack of visual details compared to daytime visible images, poses a significant challenge for human and machine interpretation. This paper proposes a novel diffusion method, dubbed Temporally Consistent Patch Diffusion Models (TC-DPM), for infrared-to-visible video translation. Our method, extending the Patch Diffusion Model, consists of two key components. Firstly, we propose a semantic-guided denoising, leveraging the strong representations of foundational models. As such, our method faithfully preserves the semantic structure of generated visible images. Secondly, we propose a novel temporal blending module to guide the denoising trajectory, ensuring the temporal consistency between consecutive frames. Experiment shows that TC-PDM outperforms state-of-the-art methods by 35.3% in FVD for infrared-to-visible video translation and by 6.1% in AP50 for day-to-night object detection. Our code is publicly available at https://github.com/dzungdoan6/tc-pdm, Comment: Technical report

Prior to deployment, an object detector is trained on a dataset compiled from a previous data collection campaign. However, the environment in which the object detector is deployed will invariably evolve, particularly in outdoor settings where changes in lighting, weather and seasons will significantly affect the appearance of the scene and target objects. It is almost impossible for all potential scenarios that the object detector may come across to be present in a finite training dataset. This necessitates continuous updates to the object detector to maintain satisfactory performance. Test-time domain adaptation techniques enable machine learning models to self-adapt based on the distributions of the testing data. However, existing methods mainly focus on fully automated adaptation, which makes sense for applications such as self-driving cars. Despite the prevalence of fully automated approaches, in some applications such as surveillance, there is usually a human operator overseeing the system's operation. We propose to involve the operator in test-time domain adaptation to raise the performance of object detection beyond what is achievable by fully automated adaptation. To reduce manual effort, the proposed method only requires the operator to provide weak labels, which are then used to guide the adaptation process. Furthermore, the proposed method can be performed in a streaming setting, where each online sample is observed only once. We show that the proposed method outperforms existing works, demonstrating a great benefit of human-in-the-loop test-time domain adaptation. Our code is publicly available at https://github.com/dzungdoan6/WSTTA

Information theory has proven to be a worthwhile tool for investigating the implications of the Higgs sector in the Next-to-minimal supersymmetric Standard Model (NMSSM) using Higgs information at the Large Hadron Collider assessed through the entropy constructed by means of the branching ratios of decay channels of the Higgs boson. The present article focuses on the parameter space of supersymmetric extension with an extra term of gauge singlet in light of various experimental constraints. Our findings show the most preferred values of $m_0$, $m_{1/2}$, $ A_0$, $ tan\beta$, $\lambda$, $\mu_{eff}$, neutralino LSP $ m_{\tilde\chi^{0}_{1}}$, lightest chargino $ m_{\tilde\chi^{\pm}_{1}}$, singlino $ m_{\tilde\chi^{0}_{5}}$, and gluino $ m_{\tilde g}$ to be around 1.93 TeV, 1.78 TeV, $-$3.62 TeV, 27.5, 0.012, 665.7 GeV, 0.74 TeV, 0.79 TeV, 11.24 TeV, and 3.70 TeV, respectively, that is compatible with the relic density of dark matter., Comment: 16 pages, 4 figures

To ensure reliable object detection in autonomous systems, the detector must be able to adapt to changes in appearance caused by environmental factors such as time of day, weather, and seasons. Continually adapting the detector to incorporate these changes is a promising solution, but it can be computationally costly. Our proposed approach is to selectively adapt the detector only when necessary, using new data that does not have the same distribution as the current training data. To this end, we investigate three popular metrics for domain gap evaluation and find that there is a correlation between the domain gap and detection accuracy. Therefore, we apply the domain gap as a criterion to decide when to adapt the detector. Our experiments show that our approach has the potential to improve the efficiency of the detector's operation in real-world scenarios, where environmental conditions change in a cyclical manner, without sacrificing the overall performance of the detector. Our code is publicly available at https://github.com/dadung/DGE-CDA., Comment: Accepted to CVIU

To see what is not in the image is one of the broader missions of computer vision. Technology to inpaint images has made significant progress with the coming of deep learning. This paper proposes a method to tackle occlusion specific to human faces. Virtual presence is a promising direction in communication and recreation for the future. However, Virtual Reality (VR) headsets occlude a significant portion of the face, hindering the photo-realistic appearance of the face in the virtual world. State-of-the-art image inpainting methods for de-occluding the eye region does not give usable results. To this end, we propose a working solution that gives usable results to tackle this problem enabling the use of the real-time photo-realistic de-occluded face of the user in VR settings., Comment: This workshop paper is presented in CVPR Workshop on Computer Vision for Augmented and Virtual Reality, New Orleans, LA, 2022. Link: https://xr.cornell.edu/workshop/2022/papers

Flavour-violating couplings of Higgs boson with stop and scharm quarks could be very important as in addition to lifting the mass of the Higgs boson by a few GeV, it could also play a vital phenomenological role in reducing the Supersymmetry breaking scale significantly. In this work, we investigate effects of such flavour-violating couplings within the Constrained Minimal Supersymmetric Standard Model (CMSSM) framework in the context of LEP data, Higgs data at the LHC, precision observables and the relic density of the dark matter using Bayesian statistics. Our detailed analysis reveals that the most probable values of $m_{0}$, $m_{1/2}$, $A_{0}$, $\tan{\beta}$, $\delta^{LR}_{ct}$ are expected to be around 4.83 TeV, 2.54 TeV, 1.90 TeV, 41.5, and 6.1$\times10^{-2}$, respectively, with flat priors. The corresponding values translate into 3.25 TeV, 2.13 TeV, 1.90 TeV, 44.7, and 5.9$\times10^{-2}$, respectively, if the natural priors are used. Furthermore, a comparison of our model with the CMSSM of flavour-conservation as the base model yields a Bayes factor of about 6 while taking into account all the experimental constraints used in our study. Our analysis also reflects that the lightest neutralino would have a mass of about 1 TeV., Comment: 17 pages, 7 figures

Information theory turns out to be an interesting tool for studying the consequences of Higgs observations to various new physics candidate theories by means of the information measure as the entropy of Higgs-Boson through its various detection modes at the Large Hadron Collider. The present article investigates the parameter space of a supersymmetric scenario where sfermions and one of the Higgs superfields are decoupled, while the gauginos, Higgsinos, and the remaining Higgs doublet are still allowed to be lighter. Our analysis reveals that this is quite a viable choice in the light of LHC discovery of a Higgs which resembles the SM Higgs-Boson and nothing else so far. While the supersymmetry breaking scale $M_S$ could be as high as $10^{11}$ GeV or so, the most preferred values of the $M_{S}$ and $\tan\beta$ are found to be around 3.6$\times10^7$ GeV and $41$ respectively, which is also consistent with the relic abundance of the neutralino dark matter. The corresponding value of neutralino ($ m_{\tilde\chi^{0}_{1}}$) LSP is estimated to be around 1.01 TeV. The preferred values of other parameters, namely, the Higgsino mass ($\mu$) and gaugino mass parameters ($M_1$ and $M_2$) are found to be about 1.05 TeV, 1.74 TeV, and 2.57 TeV, respectively., Comment: 13 pages, 3 figures, 3 tables

Flavour-violating interactions of the stop-quarks are expected to provide an additional few GeV contributions to the Higgs-Boson mass, particularly when mix with scharm-quarks, thereby allowing reduced supersymmetry (SUSY) breaking scale compared to flavour-conserving constrained minimal supersymmetric Standard Model (CMSSM). Inspired by this, we analyse the interactions mentioned above in the context of CMSSM using the information entropy of the Higgs-Boson for a wider region of flavour-violating CMSSM parameter space $(m_0,m_{1/2},A_0,$ $\tan\beta,sgn(\mu),\delta_{ct}^{ij})$, where $\delta_{ct}^{ij}$ represents the flavour-violating coupling of the top-quark with the charm-quark and $i, j$ defining left and right chiralities of squarks. Our information-theoretic analysis of the model mentioned above reveals the most favourable values of $(m_0, m_{1/2},A_0, \tan\beta,\delta_{ct}^{ij})$ as $(4.30 {\rm~TeV}, 2.32 {\rm~TeV}, -4.96{\rm~ TeV},$ $ 22.8,0.037)$ and $(4.16 {\rm ~TeV},$ $3.89~{\rm TeV},-4.10~{\rm TeV}, 19.4, 0.039)$ for $(i, j) = (L, R)$ and $(R, L)$, respectively, corresponding to the maximum entropy which suggest the SUSY breaking scale to be about $5$ TeV, thereby allowing considerable low values of sparticles masses than the flavour-conserving CMSSM., Comment: 17 pages, 8 figures

Traditionally, video conferencing is a widely adopted solution for telecommunication, but a lack of immersiveness comes inherently due to the 2D nature of facial representation. The integration of Virtual Reality (VR) in a communication/telepresence system through Head Mounted Displays (HMDs) promises to provide users a much better immersive experience. However, HMDs cause hindrance by blocking the facial appearance and expressions of the user. To overcome these issues, we propose a novel attention-enabled encoder-decoder architecture for HMD de-occlusion. We also propose to train our person-specific model using short videos (1-2 minutes) of the user, captured in varying appearances, and demonstrated generalization to unseen poses and appearances of the user. We report superior qualitative and quantitative results over state-of-the-art methods. We also present applications of this approach to hybrid video teleconferencing using existing animation and 3D face reconstruction pipelines.

We discuss information theory as a tool to investigate constrained minimal supersymmetric Standard Model (CMSSM) in the light of observation of Higgs boson at the Large Hadron Collider. The entropy of the Higgs boson using its various detection modes has been constructed as a measure of the information and has been utilized to explore a wide range of CMSSM parameter space after including various experimental constraints from the LEP data, B-physics, electroweak precision observables and relic density of dark matter. According to our study while the lightest neutralino is preferred to have a mass around 1.92 TeV, the gluino mass is estimated to be around 7.44 TeV. The values of CMSSM parameters $m_0$, $m_{1/2}$, $A_0$ and $\tan\beta$ correspond to the most preferred scenario are found to be about 6 TeV, 3.6 TeV, $-$6.9 TeV and 36.8 respectively., Comment: 16 pages, 7 figures, 3 tables

Loops that iterate over SQL query results are quite common, both in application programs that run outside the DBMS, as well as User Defined Functions (UDFs) and stored procedures that run within the DBMS. It can be argued that set-oriented operations are more efficient and should be preferred over iteration; but from real-world use cases, it is clear that loops over query results are inevitable in many situations, and are preferred by many users. Such loops, known as cursor loops, come with huge trade-offs and overheads w.r.t. performance, resource consumption and concurrency. We present Aggify, a technique for optimizing loops over query results that overcomes all these overheads. It achieves this by automatically generating custom aggregates that are equivalent in semantics to the loop. Thereby, Aggify completely eliminates the loop by rewriting the query to use this generated aggregate. This technique has several advantages such as: (i) pipelining of entire cursor loop operations instead of materialization, (ii) pushing down loop computation from the application layer into the DBMS, closer to the data, (iii) leveraging existing work on optimization of aggregate functions, resulting in efficient query plans. We describe the technique underlying Aggify and present our experimental evaluation over benchmarks as well as real workloads that demonstrate the significant benefits of this technique.

Finding community structures in social networks is considered to be a challenging task as many of the proposed algorithms are computationally expensive and does not scale well for large graphs. Most of the community detection algorithms proposed till date are unsuitable for applications that would require detection of communities in real-time, especially for massive networks. The Louvain method, which uses modularity maximization to detect clusters, is usually considered to be one of the fastest community detection algorithms even without any provable bound on its running time. We propose a novel graph traversal-based community detection framework, which not only runs faster than the Louvain method but also generates clusters of better quality for most of the benchmark datasets. We show that our algorithms run in O(|V | + |E|) time to create an initial cover before using modularity maximization to get the final cover. Keywords - community detection; Influenced Neighbor Score; brokers; community nodes; communities, Comment: 29 pages, 9 tables, and 13 figures. Accepted in "Knowledge and Information Systems", 2018