Plant tolerance, a significant parameter against saline conditions has become challenging and a hallmark for agriculture crops. There is an urgent need to incorporate green nanostructure approaches to cope with stress conditions to promote plant growth. Nanotechnology has arisen as an auspicious tool to alleviate the detrimental effects of salt stress(NaCl) in Solanum Lycopersicum via ZnO (NPs). The focal drive of this study was to prepare bioengineered zinc oxide Nanoparticles (NPs), and their characterization and evaluate different parameters in S. Lycopersicum in salinity stress conditions. Administering 50 ppm of ZnO-NPs enhanced tomato morphological characteristics; the treatments with the highest growth were the control group and 50 ppm ZnO NP treatments. Salt stress significantly reduced growth, whereas ZnO-NPs somewhat offset their effects. The biochemical data obtained showed a reduction in total proline content of 21
Image defogging is a challenging and hot-spot issue in the field of computer vision. Existing learning methods usually employ a single convolutional neural network (CNN) model to address it. However, such methods often overlook the restoration of edge details and exhibit poor defogging performance under non-uniform haze conditions. To solve the above two problems, this paper proposes a dual-branch defogging generative adversarial network that incorporates attention perception and contrastive learning. (1) The Residual Attention Branch (RAB) aims to generate attention feature maps, and the Scene Reconstruction Branch (SRB) is used to reconstruct haze-free images. (2) Considering that edge texture details may be lost in defogged images, an Attention-Aware Corrector (AAC) is proposed. (3) A Multi-Scale Fusion Discriminator (MFD) is used to supervise the restoration of haze-free images. (4) Contrastive learning is introduced as a loss function in network training, which further improves the quality of restored images. Experimental results show that the method proposed in this paper outperforms relevant classical methods on both synthetic and real-world datasets, providing new ideas and a technical baseline for image defogging.
To effectively address issues of poor efficiency and high subjectivity in manual rock core integrity assessment, a deep semantic segmentation-based intelligent prediction algorithm named DSS-RCI is proposed. In DSS-RCI, a feature extraction network based on position-aware circular convolution is first designed to capture intricate rock core details and global contextual information, significantly enhancing the network’s target localization capability and resistance to environmental interference. Subsequently, a multi-level feature enhancement network based on a feature refinement fusion network and spatial context-aware module is built to achieve refined processing and effective information enhancement of features at different levels, eliminating interference from redundant features and improving the network’s contextual feature extraction capabilities. After that, DySample, a dynamic up-sampler feature decoding module, is used to decode the enhanced feature layer to output the segmentation image of the rock core block. Finally, Rock Quality Designation (RQD) of the rock core is automatically calculated based on the segmentation results, and the rock core integrity grade is predicted. The experimental results show that the mAP and mIoU of DSS-RCI are 93.12
Methylene blue (MB) residues in aquaculture products pose significant food safety risks due to their toxicity, necessitating rapid and precise detection methods. Conventional chemical analysis methods for MB residues are often cumbersome and time consuming. This study addresses these limitations by developing a rapid, accurate, and non-destructive method for MB residue detection in perch, utilizing visible–near infrared (Vis–NIR) spectroscopy combined with advanced machine learning. Standard normal variate (SNV) was identified as the optimal spectral preprocessing technique. Subsequently, a novel model integrating SNV, monarch butterfly optimization (MBO), and a gated recurrent unit (GRU) network was established, which significantly enhanced the predictive accuracy of the GRU through optimized hyperparameters. The proposed SNV–MBO–GRU model achieved a validation R2 of 0.930, outperforming the particle swarm optimization (PSO)–optimized SNV–GRU model by substantially reducing the root mean square error (RMSE), mean squared error (MSE), and mean absolute error (MAE) by 16.1
Convolutional neural networks (CNNs) have attracted extensive attention in food quality analysis, owing to their outstanding ability to process multi-dimensional food quality data. This review summarizes the research progress and potential development trends of 1D-CNNs, 2D-CNNs, and 3D-CNNs in food quality evaluation, with a specific focus on their applications in three key data types: spectral data, image data, and spectral-spatial fused information. Nevertheless, the application of CNNs in food quality analysis still faces several persistent challenges, such as issues related to data quality, high model complexity coupled with poor interpretability, substantial computational costs, and inadequate model generalization. This review can shed new insights for promoting the wider adoption of CNNs in the food industry, and further drive the development of more intelligent and sustainable food quality perception systems.
