Images captured under low-light environments typically have poor visibility, affecting many advanced computer vision tasks. In recent years, there been some image enhancement models based on deep learning, but they not able to effectively mine the multiscale features in image, resulting generalization performance and instability of model. The disadvantages are mainly reflected color distortion, unsaturation artifacts. Current methods unable adjust exposure effectively, uneven or partial...

Image gradients contain crucial information in the images. However, gradient of low-light images is often concealed darkness and susceptible to noise contamination. This imprecise poses a significant obstacle Low-Light Enhancement (LLIE) tasks. Simultaneously, methods relying solely on pixel-level reconstruction loss struggle accurately correct mapping from dimly lit normal images, resulting restored outcomes with color abnormalities or artifacts. In this paper, we propose Gradient-Aware...

In this article, we propose a multiscale cross-connected dehazing network with scene depth fusion. We focus on the correlation between hazy image and corresponding image. The model encodes decodes separately includes cross connections at decoding end to directly generate clean in an end-to-end manner. Specifically, first construct input pyramid obtain receptive fields of multiple levels. Then, add features dimensions encoder. Finally, two paths decoder are cross-connected. addition, proposed...

Recently, large language models (LLMs) have been deployed to tackle various software engineering (SE) tasks like code generation, significantly advancing the automation of SE tasks. However, assessing quality these LLM-generated and text remains challenging. The commonly used Pass@k metric necessitates extensive unit tests configured environments, demands a high labor cost, is not suitable for evaluating text. Conventional metrics BLEU, which measure only lexical rather than semantic...

Nitrate is a critical limiting nutrient that significantly influences marine primary productivity and carbon sequestration. However, three-dimensional observation reconstruction of oceanic nitrate remain constrained by the scarcity in-situ data limited spatial coverage. To address challenge observational labels hindering development global deep learning models for estimation, this study proposes novel framework utilizes underwater signals label augmentation, thereby reducing uncertainty in...

With the rapid development of Information and Communication Technology, growth data is set to explode brings us an era cloud computing. As a result, more services are posted on web. In such situation, searching composing become computationally demanding work. To address this problem, we design system which combines fuzzy logic with GraphPlan algorithm. Fuzzy rules employed evaluate rank according users' preferences, then, solution optimal QoS value selected used in construction. We our...

In this paper, we propose an Iterative Residual Network. By designing the calculation unit, put hazy image into unit to perform iterative operation that can stitch in stages with output and substitute it calculation. After multiple iterations, a clean be generated. We introduce Long Short-Term Memory network ideas design process of computing further optimize model. The used connect units at different stages. use residual block connection deep processing preserve original features prevent...

Increasing emphasis on users' preferences and the growth of services web make service composition a time consuming complicated work. In this paper, an approach for that combines overall QoS modified graphplan is proposed. Fuzzy logic applied to handle uncertainty support decision making. Specifically, values are generated by using fuzzy analytical hierarchy process, entropy user's preferences. The pruning according in forward backward expansions, while producing solutions compositions....

Web service composition enables the provision/reusing of existing services in different business processes to satisfy requirements without investing new infrastructure. QoS-aware web seeks help users find optimal solution with maximization users' satisfaction. A number approaches based on Communicating Sequential Processes (CSP) have been proposed model and verify properties composition. However, little work has done verifying inputs, outputs QoS criteria In this paper, we present a...

Low-light remote sensing images generally feature high resolution and spatial complexity, with continuously distributed surface features in space. This continuity scenes leads to extensive long-range correlations domains within images. Convolutional Neural Networks, which rely on local for long-distance modeling, struggle establish such On the other hand, transformer-based methods that focus global information face computational complexities when processing high-resolution From another...

Underwater images often suffer from various issues such as low brightness, color shift, blurred details, and noise due to light absorption scattering caused by water suspended particles. Previous underwater image enhancement (UIE) methods have primarily focused on spatial domain enhancement, neglecting the frequency information inherent in images. However, degradation factors of are closely intertwined domain. Although certain focus enhancing domain, they overlook relationship between...