In recent years, jute, as an important natural fiber crop, has become more and more significant in the production process of insect pests, causing serious harm to agricultural production. Especially in the field of crop pest identification with complex backgrounds, fuzzy features, and multiple small targets, the lack of datasets specifically for jute pests has led to the large limitations of traditional pest identification models in terms of generalization. At the same time, the research on models specifically for jute pest detection is still in its infancy. To solve this problem, we constructed a large-scale image dataset containing nine types of jute pests, which was highly targeted and could effectively support model training and evaluation. In this study, we developed a deep convolutional neural network model based on YOLOv7, namely JutePest-YOLO. The model has optimized the Backbone, Head, and loss functions of the baseline model, and introduced the new ELAN-P module and P6 detection layer, which effectively improved the model’s ability to identify jute pests in complex backgrounds. The experimental results showed that compared with the baseline model, the Precision, Recall, and F1 scores of the JutePest-YOLO model were improved by 3.45%, 1.76%, and 2.58%, respectively; the mAP@0.5 and mAP@0.5:0.95 was improved by 2.24% and 3.25%, and the overall model’s computation (GFLOPS) was reduced by 16.05%. Compared to other advanced methods such as YOLOv8s, JutePest-YOLO has achieved superior performance in terms of detection accuracy, with a precision of 98.7% and mAP@0.5 reaching 95.68%. As a result, JutePest-YOLO not only achieved significant improvement in recognition accuracy but also optimized computational efficiency. It’s a high-performance, lightweight solution for jute pest detection.

The rapid changes in the global environment have led to an unprecedented decline in biodiversity, with over 28% of species facing extinction. This includes snakes, which are key to ecological balance. Detecting snakes is challenging due to their camouflage and elusive nature, causing data loss and feature extraction difficulties in ecological monitoring. To address these challenges, we propose an enhanced snake detection model, Snake-DETR, based on RT-DETR, specifically designed for snake detection in complex natural environments. First, we designed the Enhanced Generalized Efficient Layer Aggregation Network Based on Context Anchor Attention, which enhances the feature extraction capability for occluded snakes by aggregating critical layer information and strengthening context-dependent feature extraction. Additionally, we introduced the Enhanced Feature Extraction Backbone Network Based on Context Anchor Attention, which manages input information using multiple Enhanced Generalized Efficient Layer Aggregation Networks to retain essential spatial and semantic information. Subsequently, a lightweight Group-Shuffle Convolution is used to optimize the encoder, which reduces dependency on large-scale training data, thereby making it suitable for deployment on edge devices. Finally, we incorporated the Powerful-IoU loss function to improve regression path accuracy. Experimental results on a custom dataset covering 27 snake species demonstrate that Snake-DETR achieves a good balance between model efficiency and detection performance, meeting the requirements for fine-grained snake object detection. Compared to other state-of-the-art models, Snake-DETR achieved an accuracy of 97.66%, a recall rate of 93.92%, mAP@0.5 of 95.23%, and mAP@0.5:0.95 of 72.15%, all outperforming other algorithms in the comparative tests. Furthermore, the computational load and parameter count of the model are reduced by 47.2 and 52.2%, respectively, compared to the benchmark model. Additionally, the real-time processing capability is 43.5 frames per second, meeting the demand for real-time processing. Snake-DETR demonstrates excellent performance in complex environments and is suitable for wild snake fauna monitoring and edge device deployment, providing key technical support for ecological research.

In the field of abstract text summarization, architectures based on encoder-decoder frameworks are widely applied to sequence-to-sequence generation tasks and can effectively handle sequences of unlimited length. Subsequently, the transformer model use a global attention mechanism, allowing encodings at different distances to mutually interact, greatly enhancing the model’s contextual awareness. However, this context-awareness is global, requiring the model to additionally learn to extract different levels of information to increase understanding. We improve the structure of the model to introduce prior knowledge so that it can learn from the global and local information and enhance the model’s understanding ability. This paper proposes global information-aware encoding and local information-aware encoding, which enhance the understanding of documents from coarse-grained and fine-grained perspectives respectively. Global encoding adds an extra feature to the encoder stage and performs attention with the document, generating a global summary encoding of the entire document to guide the generation of the summary content. Local encoding is to perform local convolution on the features extracted by the encoder, use prior knowledge to extract local features of the document and enable the model to quickly extract local detail information. Experiments show that the improved model proposed in this paper has higher rouge scores than the baseline model on the LCSTS and CSL datasets, and also has advantages over some mainstream models. The generated summaries are more accurate and informative. The code is available on github.

Three-dimensional gravity inversion technology involves inferring the underground density structure based on observed gravity anomaly data. In addition to gravity inversion based on physics-driven methods, deep learning, as a purely data-driven technique, is increasingly gaining attention in geophysical inversion problems. However, purely data-driven methods rely on the implicit relationships within the data during the inversion process, which results in a lack of clear physical significance. This study proposes a three-dimensional gravity inversion method that integrates physical equations with deep learning. Based on the U-Net architecture, the gravity forward equation is incorporated as a physical constraint term, and a composite loss function—comprising three-dimensional mean squared error, a depth-weighting function, and three-dimensional intersection-over-union loss—is constructed to enhance inversion accuracy. Numerical experiments indicate that this method outperforms traditional algorithms in terms of density recovery accuracy and boundary clarity. When applied to gravity anomaly data from the Tangshan earthquake region in China, this method successfully inverted the three-dimensional subsurface density structure, revealing a high-density anomaly beneath the seismic source area, which provides important evidence for understanding the regional earthquake generation mechanism.

Background: Knee osteoarthritis (KOA) is a degenerative joint disease commonly assessed using X-ray images based on the Kellgren-Lawrence (KL) criteria. Although the KL standard exists, its ambiguity often causes patients to misunderstand their condition, leading to overtreatment or delayed treatment and challenges in guiding precise surgical decisions. Moreover, the data-driven technology has been impeded by low resolution and feature distribution inconsistency of knee X-ray images. The imbalances between positive and negative samples further degrade detection accuracy. Objective: The objective of this study was to develop a deep learning-based model, namely Task-aligned Path Aggregation Feature Fusion For Knee Osteoarthritis Detection (TPAFFKnee), to improve KOA detection accuracy by addressing limitations in traditional methods. Its more accurate detection could help in terms of proper treatment for patients and precision in surgery by physicians. Methods: We proposed the TPAFFKnee model based on the EfficientNetB4 network, which introduced a path aggregation network for better feature extraction and replaced Fully Convolutional Network (FCN) with task-aligned detection as the head. Additionally, the loss function was improved by replacing the original loss function with Efficient Intersection over Union Loss (EIoU Loss) to address the imbalance between positive and negative samples. Results: The results showed that the model could accurately detect KOA categories and lesion locations based on the KL classification criteria, with a Mean Average Precision (mAP) of 93% on the Mendeley KOA dataset of 1650 knee osteoarthritis X-ray images from several hospitals. The mAP for the K2, K3, and K4 categories were 98.6%, 98.5%, and 99.6%, respectively. Compared with Faster R-CNN, SSD, RetinaNet, EfficientNetB4, and YOLOX, the proposed algorithm improved detection mAP by 14.3%, 12.4%, 15.3%, 22.7%, and 4.3%. Conclusion: This study emphasizes the importance of the EfficientNetB4 network in KOA detection. The TPAFFKnee model provides an effective solution for improving the accuracy of KOA detection and offers a promising approach for standardized KL classification in medical applications. Future research can integrate more clinical data while improving the overall landscape of healthcare delivery through data-driven automation solutions.