An official website of the United States government
The performance of object detection models in adverse weather conditions remains a critical challenge for intelligent transportation systems. Since advancements in autonomous driving rely heavily on extensive datasets, which help autonomous driving systems be reliable in complex driving environments, this study provides a comprehensive dataset under diverse weather scenarios like rain, haze, nighttime, or sun flares and systematically evaluates the robustness of state-of-the-art deep learning-based object detection frameworks. Our Adverse Driving Conditions Dataset features eight single weather effects and four challenging mixed weather effects, with a curated collection of 50,000 traffic images for each weather effect. State-of-the-art object detection models are evaluated using standard metrics, including precision, recall, and IoU. Our findings reveal significant performance degradation under adverse conditions compared to clear weather, highlighting common issues such as misclassification and false positives. For example, scenarios like haze combined with rain cause frequent detection failures, highlighting the limitations of current algorithms. Through comprehensive performance analysis, we provide critical insights into model vulnerabilities and propose directions for developing weather-resilient object detection systems. This work contributes to advancing robust computer vision technologies for safer and more reliable transportation in unpredictable real-world environments.
more »
« less
This content will become publicly available on December 1, 2027
Full vision adaptation in mixed-light conditions enabled by dynamic water adsorption/desorption
Abstract Mimicking the human eye’s ability to autonomously adapt to diverse and mixed illumination conditions remains a fundamental challenge in artificial vision systems. Although substantial progress has been made in materials and device engineering, current adaptive vision architectures still depend heavily on complex circuitry or algorithms and are typically restricted to uniform illumination owing to the strong intensity-dependence of photosensitivity. Here, this work presents a highly adaptive TiO₂/PEDOT:PSS photomemristor that leverages the tunable conductivity of PEDOT:PSS together with the optoelectronic response of TiO₂. The photothermal effect dynamically modulates the water absorption/desorption equilibrium in PEDOT:PSS, enabling reversible suppression or enhancement of photosensitivity under bright or dim illumination, respectively. By combining with artificial neural networks (ANNs), the artificial vision system based on TiO₂/PEDOT:PSS photomemristor arrays achieves a high accuracy of 91.3% in image recognition under mixed-light conditions—without the need for complex circuitry or algorithms. This work may establish a new approach for designing autonomous, efficient, and high-performance neuromorphic vision systems to advance the development of autonomous driving and humanoid robots.
more »
« less
- PAR ID:
- 10694990
- Publisher / Repository:
- Nature Springer
- Date Published:
- Journal Name:
- Nature Communications
- Volume:
- 17
- Issue:
- 1
- ISSN:
- 2041-1723
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract Titanium dioxide (TiO 2 ) has been used in numerous paintings since its creation in the early 1920s. However, due to this relatively recent adoption by the art world, we have limited knowledge about the nature and risk of degradation in museum environments. This study expands on the existing understanding of TiO 2 facilitated degradation of linseed oil, by examining the effect of visible light and crystallographic phase (either anatase or rutile) on the reactivity of TiO 2 . The present approach is based on a combination of experimental chemical characterization with computational calculation through Density Functional Theory (DFT) modeling of the TiO 2 -oil system. Attenuated Total Reflection Fourier Transform Infrared Spectroscopy (ATR-FT-IR) enabled the identification of characteristic degradation products during UV and visible light aging of both rutile and anatase based paints in comparison to BaSO 4 and linseed oil controls. In addition, cratering and cracking of the paint surface in TiO 2 based paints, aged under visible and UV–vis illumination, were observed through Scanning Electron Microscopy (SEM). Finally, Density Functional Theory (DFT) modeling of interactions between anatase TiO 2 and oleic acid, a fatty acid component of linseed oil, to form a charge transfer complex explains one possible mechanism for the visible light activity observed in artificial aging. Visible light excitation of this complex sensitizes TiO 2 by injecting an electron into the conduction band of TiO 2 to generate reactive oxygen species and subsequent degradation of the oil binder by various mechanisms (e.g., formation of an oleic acid cation radical and other oxidation products). Graphical Abstractmore » « less
-
Abstract Nonradiating optical anapoles are special configurations of charge‐current distributions that do not radiate. It was theoretically predicted that, for microspheres, electric and magnetic dipolar coefficients can simultaneously vanish by engineering the incident light, leading to the excitation of nonradiatinghybridoptical anapoles. In this work, the experimental detection of hybrid optical anapoles in dielectric microspheres (TiO2) is reported using dual detection optical spectroscopy, developed to enable sequential measurement of forward and backward scattering under tightly‐focused Gaussian beam (TFGB) illumination. The results show that the excitation of TiO2microspheres (diameter,d≈1 µm) under TFGB illumination leads to the appearance of scattering minima in both the forward and backward directions within specific wavelength ranges. These scattering minima are found to be due to vanishing electric and magnetic dipolar coefficients associated with hybrid optical anapoles. The ability to confine electromagnetic fields associated with hybrid optical anapoles can give rise to several novel optical phenomena and applications.more » « less
-
Agaian, Sos S.; Jassim, Sabah A. (Ed.)Face recognition technologies have been in high demand in the past few decades due to the increase in human-computer interactions. It is also one of the essential components in interpreting human emotions, intentions, facial expressions for smart environments. This non-intrusive biometric authentication system relies on identifying unique facial features and pairing alike structures for identification and recognition. Application areas of facial recognition systems include homeland and border security, identification for law enforcement, access control to secure networks, authentication for online banking and video surveillance. While it is easy for humans to recognize faces under varying illumination conditions, it is still a challenging task in computer vision. Non-uniform illumination and uncontrolled operating environments can impair the performance of visual-spectrum based recognition systems. To address these difficulties, a novel Anisotropic Gradient Facial Recognition (AGFR) system that is capable of autonomous thermal infrared to visible face recognition is proposed. The main contribution of this paper includes a framework for thermal/fused-thermal-visible to visible face recognition system and a novel human-visual-system inspired thermal-visible image fusion technique. Extensive computer simulations using CARL, IRIS, AT&T, Yale and Yale-B databases demonstrate the efficiency, accuracy, and robustness of the AGFR system. Keywords: Infrared thermal to visible facial recognition, anisotropic gradient, visible-to-visible face recognition, nonuniform illumination face recognition, thermal and visible face fusion methodmore » « less
-
Artificial intelligence‐augmented additive manufacturing (AI2AM) represents a transformative frontier in digital fabrication, where artificial intelligence (AI) is embedded not as a peripheral tool, but as a central framework driving intelligent, adaptive, and autonomous additive manufacturing (AM) systems. The objective of this Roadmap is to present a comprehensive vision of the state‐of‐the‐art developments in AI2AM while charting the future trajectory of this rapidly emerging field. As AM applications continue to expand across diverse sectors, conventional design and control strategies face growing limitations in scalability, quality assurance, and material complexity. AI uses tools like computer vision, generative design, and large language models to help solve problems in scalability, quality assurance, and material complexity, allowing for real‐time defect detection, digital twin integration, and closed‐loop process control. This roadmap brings together leading contributions from twenty internationally recognized research groups by uniting perspectives from materials science, computer science, robotics, and manufacturing. This work aims to create a cohesive framework for advancing AI2AM as a multidisciplinary science. The ultimate intent of this work is to establish a foundation for coordinated research and innovation in AI‐powered AM and to serve as a strategic entry point for future breakthroughs in autonomous and sustainable production.more » « less
