Search

Adverse conditions like snow, rain, nighttime, and fog, pose challenges for autonomous driving perception systems. Existing methods have limited effectiveness in improving essential computer vision tasks, such as semantic segmentation, and often focus on only one specific condition, such as removing rain or translating nighttime images into daytime ones. To address these limitations, we propose a method to improve the visual quality and clarity degraded by such adverse conditions. Our method, AllWeather-Net, utilizes a novel hierarchical architecture to enhance images across all adverse conditions. This architecture incorporates information at three semantic levels: scene, object, and texture, by discriminating patches at each level. Furthermore, we introduce a Scaled Illumination-aware Attention Mechanism (SIAM) that guides the learning towards road elements critical for autonomous driving perception. SIAM exhibits robustness, remaining unaffected by changes in weather conditions or environmental scenes. AllWeather-Net effectively transforms images into normal weather and daytime scenes, demonstrating superior image enhancement results and subsequently enhancing the performance of semantic segmentation, with up to a 5.3% improvement in mIoU in the trained domain. We also show our model's generalization ability by applying it to unseen domains without re-training, achieving up to 3.9% mIoU improvement. Code can be accessed at: https://github.com/Jumponthemoon/AllWeatherNet.

This study comprehensively examined the structural, electronic, electrochemical, and energy storage properties of boron-vacancy induced porous boron nitride monolayers (BN:VB) as multifunctional materials, anodes for MIBs and H2 storage applications. Our computational approaches, density functional theory (DFT), ab initio molecular dynamics (AIMD), and thermodynamic analysis, revealed exceptionally high energy and gravimetric densities for MIBs and H2 storage, respectively. We investigated the interactions of Li, Na, and K atoms on BN:VB, which strongly bonded with binding energies stronger than their bulk cohesive energies, which ensured structural stability and the absence of metal clustering. Electronic properties, analyzed through spin-polarized partial density of states (PDOS), band structure, and Bader charge analysis, revealed significant charge transfers from the metal atoms to BN:VB, enhancing the electronic conductivity of the latter. Theoretical specific capacities were calculated as 1821.53, 786.11, and 490.51 mA h/g for Li, Na, and K, respectively, which comfortably exceeded the conventional anodes, such as graphite. Average open-circuit voltages (OCVs) were found as 0.15, 0.25, and 0.32 V, for Li, Na, and K, respectively, indicating strong electrochemical stability. Diffusion studies showed lower barriers of 0.47, 0.08, and 0.60 eV for Li, Na, and K, respectively, with increased metal loadings, suggesting enhanced mobilities and charge/discharge rates. On the other side, the metal-functionalized BN:VB monolayers exhibited remarkably high H2 gravimetric capacities, supported by Langmuir adsorption model-based statistical thermodynamic analysis. Average adsorption energies of H2 on 4Li-, 4Na-, and 4K@BN:VB, were found in perfect range for practical storage applications.

Exceptionally high energy density by mass, natural abundance, widespread applications, and environmental friendliness make hydrogen (H2) a front-runner among clean energy options. However, the transition toward clean and renewable energy applications and the actualization of H2 economy require an efficient H2 storage medium. Material-based H2 storage is a viable option, as liquefaction and storage under pressure require ultra-low temperature (-253{\deg}C) and tremendously high pressure (700 atm), respectively. In this work, we highlight the exceptional H2 storage capabilities of recently synthesized boron monoxide (BO) monolayer functionalized with light metals (Li, Na, K, and Ca). Our computational approach, employing density functional theory (DFT), ab initio molecular dynamics (AIMD), and thermodynamic analysis, reveals promising results. We found that up to four metal dopants (Li, Na, K, and Ca) can be adsorbed onto BO monolayer with significantly strong binding energies. Importantly, these bindings surpass the cohesive counterparts of the parental metal bulks, consequently stabilizing the crystal integrities, as confirmed by AIMD simulations. Each metal dopant on BO efficiently adsorbs multiple H2 molecules through electrostatic and van der Waals interactions. Interestingly, the metal-functionalized BO monolayers exhibit exceptionally high H2 gravimetric capacities up to 11.75 wt%. These promising capacities exceed the 5.50 wt% target set by the US Department of Energy for 2025. Following the same guidelines, the average binding energy per H2 molecule is within the range of -0.17 to -0.32 eV. The adsorption and desorption of H2 under practical working conditions are investigated by Langmuir adsorption model based statistical thermodynamic analysis, further supporting the potential of metal-functionalized BO monolayers for material-based H2 storage applications.

The increasing integration of renewable energy resources in evolving bulk power system (BPS) is impacting the system inertia. Type-5 wind turbine generation has the potential to behave like a traditional synchronous generator and can help improve system inertia. Hydraulic torque converter (TC) and gearbox with torque limiting feature are integral parts of a Type-5 wind turbine unit. High fidelity model of Type-5 wind turbine including these core components is not openly and widely available for grid integration and transient stability studies. This hinders appropriate assessment of Type-5 wind power plant's contribution to bulk grid resilience. This work develops a TC model based on those generally used in automobile's transmission system. Moreover, the concept of torsional coupling is leveraged to integrate the TC and gearbox system dynamics. The entire integrated model will be open sourced and publicly available for grid integration studies., Comment: This manuscript is originally submitted to Journal of Physics

Autonomous vehicles (AVs) are becoming an indispensable part of future transportation. However, safety challenges and lack of reliability limit their real-world deployment. Towards boosting the appearance of AVs on the roads, the interaction of AVs with pedestrians including "prediction of the pedestrian crossing intention" deserves extensive research. This is a highly challenging task as involves multiple non-linear parameters. In this direction, we extract and analyse spatio-temporal visual features of both pedestrian and traffic contexts. The pedestrian features include body pose and local context features that represent the pedestrian's behaviour. Additionally, to understand the global context, we utilise location, motion, and environmental information using scene parsing technology that represents the pedestrian's surroundings, and may affect the pedestrian's intention. Finally, these multi-modality features are intelligently fused for effective intention prediction learning. The experimental results of the proposed model on the JAAD dataset show a superior result on the combined AUC and F1-score compared to the state-of-the-art.

There has been budding demand for the fast, reliable, inexpensive, non-invasive, sensitive, and compact sensors with low power consumption in various fields, such as defence, chemical sensing, health care, and safe environment monitoring units. Particularly, an efficient detection of chemical warfare agents (CWAs) is of great importance for the safety and security of the humans. Inspired by this, we explored molybdenum carbide MXenes (Mo2CTx; Tx= O, F, S) as efficient sensors towards selected CWAs, such as arsine (AsH3), mustard gas (C4H8Cl2S), cyanogen chloride (NCCl), and phosgene (COCl2) both in aqueous and non-aqueous mediums. Our van der Waals corrected density functional theory (DFT) calculations reveal that the CWAs bind with Mo2CF2, and Mo2CS2 monolayers under strong chemisorption with binding energies in the range of -2.33 to -4.05 eV, whereas Mo2CO2 results in comparatively weak bindings of -0.29 to -0.58 eV. We further report the variations in the electronic properties, electrostatic potentials and work functions of Mo2CTx upon the adsorption of CWAs, which authenticate an efficient sensing mechanism. Statistical thermodynamic analysis is applied to explore the sensing properties of Mo2CTx at various of temperatures and pressures. We believe that our findings will pave the way to an innovative class of low-cost reusable sensors for the sensitive and selective detection of highly toxic CWAs in air as well as in aqueous media.

Salient object detection (SOD) extracts meaningful contents from an input image. RGB-based SOD methods lack the complementary depth clues; hence, providing limited performance for complex scenarios. Similarly, RGB-D models process RGB and depth inputs, but the depth data availability during testing may hinder the model's practical applicability. This paper exploits only RGB images, estimates depth from RGB, and leverages the intermediate depth features. We employ a pyramidal attention structure to extract multi-level convolutional-transformer features to process initial stage representations and further enhance the subsequent ones. At each stage, the backbone transformer model produces global receptive fields and computing in parallel to attain fine-grained global predictions refined by our residual convolutional attention decoder for optimal saliency prediction. We report significantly improved performance against 21 and 40 state-of-the-art SOD methods on eight RGB and RGB-D datasets, respectively. Consequently, we present a new SOD perspective of generating RGB-D SOD without acquiring depth data during training and testing and assist RGB methods with depth clues for improved performance. The code and trained models are available at https://github.com/tanveer-hussain/EfficientSOD2, Comment: Accepted at CVPRW 2022. (2022 IEEE CVPR Workshop on Fair, Data Efficient and Trusted Computer Vision)

Salient Object Detection (SOD) domain using RGB-D data has lately emerged with some current models' adequately precise results. However, they have restrained generalization abilities and intensive computational complexity. In this paper, inspired by the best background/foreground separation abilities of deformable convolutions, we employ them in our Densely Deformable Network (DDNet) to achieve efficient SOD. The salient regions from densely deformable convolutions are further refined using transposed convolutions to optimally generate the saliency maps. Quantitative and qualitative evaluations using the recent SOD dataset against 22 competing techniques show our method's efficiency and effectiveness. We also offer evaluation using our own created cross-dataset, surveillance-SOD (S-SOD), to check the trained models' validity in terms of their applicability in diverse scenarios. The results indicate that the current models have limited generalization potentials, demanding further research in this direction. Our code and new dataset will be publicly available at https://github.com/tanveer-hussain/EfficientSOD

This letter is a proof of concept for an improved transmission switching (TS) performance by moving the search space to load shed buses. Research from the past shows that changing transmission system topology changes the power flows and removes post contingency violations. Hence, TS can reduce the amount of load shed after an N-1 contingency. One of the major challenges is to find the best TS candidate in a suitable time. In this letter, the best TS candidate is determined by using a novel heuristic bi-level method based on linear sensitivity. The proposed bi-level method is easy to implement in the real world, guarantees removal of post contingency violations, and ranks the best TS candidates based on minimum load shedding possible. Moreover, the proposed method is computationally efficient since it does not involve mixed integer programming. The bi-level method is implemented by modifying the topology of transmission system after the N-1 contingency in the IEEE 39-bus test system and results show that TS with generation re-dispatch is the best solution for load shed recovery to prevent cascading failures. Moreover, the bi-level method performs even for the case when the existing methods in literature fail to completely remove post contingency violations.