Search

Semantic segmentation forms the foundation for understanding very high resolution (VHR) remote sensing images, with extensive demand and practical application value. The convolutional neural networks (CNNs), known for their prowess in hierarchical feature representation, have dominated the field of semantic image segmentation. Recently, hierarchical vision transformers such as Swin have also shown excellent performance for semantic segmentation tasks. However, the hierarchical structure enlarges the receptive field to accumulate features and inevitably leads to the blurring of object boundaries. We introduce a novel object-aware network, Embedding deep SuperPixel, for VHR image semantic segmentation called ESPNet, which integrates advanced ConvNeXt and the learnable superpixel algorithm. Specifically, the developed task-oriented superpixel generation module can refine the results of the semantic segmentation branch by preserving object boundaries. This study reveals the capability of utilizing deep convolutional neural networks to accomplish both superpixel generation and semantic segmentation of VHR images within an integrated end-to-end framework. The proposed method achieved mIoU scores of 84.32, 90.13, and 55.73 on the Vaihingen, Potsdam, and LoveDA datasets, respectively. These results indicate that our model surpasses the current advanced methods, thus demonstrating the effectiveness of the proposed scheme. [ABSTRACT FROM AUTHOR]

In agriculture, one of the most crucial elements for sustained plant production is light. Artificial lighting can meet the specific light requirements of various plants. However, it is a challenge to find optimal lighting schemes that can facilitate a balance of plant growth and nutritional qualities. In this study, we experimented with the light intensity required for plant growth and nutrient elements. We designed three light intensity treatments, 180 μmol m−2 s−1 (L1), 210 μmol m−2 s−1 (L2), and 240 μmol m−2 s−1 (L3), to investigate the effect of light intensity on lettuce growth and quality. It can be clearly seen from the radar charts that L2 significantly affected the plant height, fresh weight, dry weight, and leaf area. L3 mainly affected the canopy diameter and root shoot ratio. The effect of L1 on lettuce phenotype was not significant compared with that of the others. The total soluble sugar, vitamin C, nitrate, and free amino acid in lettuce showed more significant increases under the L2 treatment than under the other treatments. In addition, the transpiration rate and stomatal conductance were opposite to each other. The comprehensive evaluation of the membership function value method and heatmap analysis showed that lettuce had the highest membership function value in L2 light intensity conditions, indicating that the lettuce grown under this light intensity could obtain higher yield and better quality. This study provides a new insight into finding the best environmental factors to balance plant nutrition and growth. [ABSTRACT FROM AUTHOR]

This work reports the synthesis of CuxSny alloy aerogels for electrochemical CO2 reduction catalysts. An in situ reduction and the subsequent freeze-drying process can successfully give CnxSny aerogels with tuneable Sn contents, and such aerogels are composed of three-dimensional architectures made from inter-connected fine nanoparticles with pores as the channels. Density functional theory (DFT) calculations show that the introduction of Sn in Cu aerogels inhibits H2 evolution reaction (HER) activity, while the accelerated CO desorption on the catalyst surface is found at the same time. The porous structure of aerogel also favors exposing more active sites. Counting these together, with the optimized composition of Cu95Sn5 aerogel, the high selectivity of CO can be achieved with a faradaic efficiency of over 90% in a wide potential range (−0.7 V to −1.0 V vs. RHE). [ABSTRACT FROM AUTHOR]

H2 is promising clean energy and electrochemical hydrogen evolution reaction offers a sustainable route compared with traditional methane steam reforming. In this work, highly porous Co2Ni and Co2Ni/reduced graphene oxide composite aerogels are prepared by a simple in-situ reduction-gelation method in an aqueous solution. The synthesis starts with Co-Ni hydrogel which is prepared by mixing a strong reduction agent NaBH4 with the corresponding metal ions, while graphene oxide is reduced simultaneously in the gelation process. After supercritical drying, composite aerogels composed with interconnected binary Co2Ni nanoparticles network and reduced graphene oxide sheets are obtained. Such optimized composite alloy/reduced graphene oxide aerogels show good hydrogen evolution catalytic activity in alkaline electrolytes, which is highlighted by the low on-set overpotential (143 mV@10 mA/cm2) and small Tafel slope (85 mV/dec). Highlights: 3D-porous Co2Ni/rGO composite aerogel was prepared by an in-situ reduction. rGO increased the surface area and promoted the charge transfer capability. The composite aerogel exhibited good HER catalytic activity. [ABSTRACT FROM AUTHOR]