For the routine target detection algorithm in the underwater complex environment to obtain the image of the existence of blurred images, complex background and other phenomena, leading to difficulties in model feature extraction, target miss detection and other problems. Meanwhile, an improved YOLOv7 model is proposed in order to improve the accuracy and real-time performance of the underwater target detection model. The improved model is based on the single-stage target detection model YOLOv7, incorporating the CBAM attention mechanism in the model, so that the feature information of the detection target is weighted and enhanced in the spatial dimension and the channel dimension, capturing the local relevance of feature information, making the model more focused on target feature information, improved detection accuracy, and using the SPPFCSPC module, reducing the computational effort of the model while keeping the model perceptual field unchanged, improved inference speed of the model. After a large number of comparison experiments and ablation experiments, it is proved that our proposed ACFP-YOLO algorithm model has higher detection accuracy compared with Efficientdet, Faster-RCNN, SSD, YOLOv3, YOLOv4, YOLOv5 models and the latest YOLOv7 model, and is more accurate for target detection tasks in complex underwater environments advantages.

To create a risk model of aging-related long non-coding RNAs (arlncRNAs) and determine whether they might be useful as markers for risk stratification, prognosis prediction, and targeted therapy guidance for patients with lung adenocarcinoma (LUAD). Data on aging genes and lncRNAs from LUAD patients were obtained from Human Aging Genomic Resources 3 and The Cancer Genome Atlas, and differential co-expression analysis of established differentially expressed arlncRNAs (DEarlncRNAs) was performed. They were then paired with a matrix of 0 or 1 by cyclic single pairing. The risk coefficient for each sample of LUAD individuals was obtained, and a risk model was constructed by performing univariate regression, least absolute shrinkage and selection operator regression analysis, and univariate and multivariate Cox regression analysis. Areas under the curve were calculated for the 1-, 3-, and 5-year receiver operating characteristic curves to determine Akaike information criterion-based cutoffs to identify high- and low-risk groups. The survival rate, correlation of clinical characteristics, malignant-infiltrating immune-cell expression, ICI-related gene expression, and chemotherapeutic drug sensitivity were contrasted with the high- and low-risk groups. We found that 99 DEarlncRNAs were upregulated and 12 were downregulated. Twenty pairs of DEarlncRNA pairs were used to create a prognostic model. The 1-, 3-, and 5-year survival curve areas of LUAD individuals were 0.805, 0.793, and 0.855, respectively. The cutoff value to classify patients into two groups was 0.992. The mortality rate was higher in the high-risk group. We affirmed that the LUAD outcome-related independent predictor was the risk score (p < 0.001). Validation of tumor-infiltrating immune cells and ICI-related gene expression differed substantially between the groups. The high-risk group was highly sensitive to docetaxel, erlotinib, gefitinib, and paclitaxel. Risk models constructed from arlncRNAs can be used for risk stratification in patients with LUAD and serve as prognostic markers to identify patients who might benefit from targeted and chemotherapeutic agents.

In this paper, we propose a multi-scale residual attention network (MSR-Net) segmentation algorithm, which uses the ResNet50 residual network as the backbone feature extraction network and introduces a multi-scale channel attention mechanism. The MSR-Net uses the ResNet50 residual network as the backbone feature extraction network and introduces a multi-scale channel attention mechanism, which enables the network model to retain more complete sample edge information, significantly improves the segmentation capability of the model and ensures its network performance, which can effectively meet the needs of underwater image segmentation-related tasks.The proposed network is tested on the DUT-USEG dataset, and the recall, accuracy and average cross-merge ratio are 74.17%, 83.21% and 65.96%, respectively. As shown by the experimental results, compared with the classical U-Net, PSPNet and DeepLabV3, the performance indexes of the method in this paper are significantly improved.

Ocean exploration has always been an important strategic direction for the joint efforts of all mankind. Many countries in the world today are developing their own underwater autonomous explorers to better explore the seabed. Vision, as the core technology of autonomous underwater explorers, has a great impact on the efficiency of exploration. Different from traditional tasks, the lack of ambient light on the seabed makes the visual system more demanding. In addition, the complex terrain on the seabed and various creatures with different shapes and colors also make exploration tasks more difficult. In order to effectively solve the above problems, we combined the traditional models to modify the structure and proposed an algorithm for the super-resolution fusion of enhanced extraction features to perform semantic segmentation of seabed scenes. By using a structurally reparameterized backbone network to better extract target features in complex environments, and using subpixel super-resolution to combine multiscale feature semantic information, we can achieve superior ocean scene segmentation performance. In this study, multiclass segmentation and two-class segmentation tests were performed on the public datasets SUIM and DeepFish, respectively. The test results show that the mIoU and mPA indicators of our proposed method on SUIM reach 84.52% and 92.33%mPA, respectively. The mIoU and mPA on DeepFish reach 95.26% and 97.38%, respectively, and the proposed model achieves SOTA compared with state-of-the-art methods. The proposed model and code are exposed via Github1.