Extraction Ability Research Articles

Evaluating Antioxidants Present in Grapefruit Peel Phytochemicals Using Blending Extraction Method

Grapefruit peels, often discarded as waste, have gained attention for their potential as a source of natural antioxidants, addressing the growing demand for sustainable and cost-effective alternatives to synthetic compounds. Oxidative stress-related disorders continue to pose significant health challenges, emphasizing the need for natural antioxidant sources. This study investigates the antioxidant properties of grapefruit peel phytochemicals using the blending extraction method. Fresh grapefruits were purchased from Minna main market, Niger State Nigeria and stored under controlled conditions at room temperature before further analysis. The extracted phytochemicals were analyzed for their antioxidant activity and phytochemical composition. The findings reveal a concentration-dependent increase in the extract's ability to inhibit lipid peroxidation and scavenge DPPH free radicals, highlighting its strong antioxidant potential. Phytochemical analysis indicates a significant presence of phenolic compounds, while alkaloids and flavonoids were detected in lower quantities. Notably, saponins were found in higher concentrations among the analyzed phytochemicals. These results emphasize the potential of grapefruit peel extract as a rich source of natural antioxidants with promising applications in health, nutrition, and functional food development.

Sequential Multimodal Underwater Single-Photon Lidar Adaptive Target Reconstruction Algorithm Based on Spatiotemporal Sequence Fusion

For the demand for long-range and high-resolution target reconstruction of slow-moving small underwater targets, research on single-photon lidar target reconstruction technology is being carried out. This paper reports the sequential multimodal underwater single-photon lidar adaptive target reconstruction algorithm based on spatiotemporal sequence fusion, which has strong information extraction and noise filtering ability and can reconstruct the target depth and reflective intensity information from complex echo photon time counts and spatial pixel relationships. The method consists of three steps: data preprocessing, sequence-optimized extreme value inference filtering, and collaborative variation strategy for image optimization to achieve high-quality target reconstruction in complex underwater environments. Simulation and test results show that the target reconstruction method outperforms the current imaging algorithms, and the built single-photon lidar system achieves underwater lateral and distance resolution of 5 mm and 2.5cm@6AL, respectively. This indicates that the method has a great advantage in sparse photon counting imaging and possesses the capability of underwater target imaging under the background of strong light noise. It also provides a good solution for underwater target imaging of small slow-moving targets with long-distance and high-resolution.

A Target Domain-Specific Classifier Weight Partial Transfer Adversarial Network for Bearing Fault Diagnosis

In actual industry applications, the failure categories of practical equipment are usually a subset of laboratory conditions failure categories. Due to the strict constraints, partial transfer learning can address a more practical diagnostic scenario. In view of this, this paper proposes a target domain-specific classifier weight partial transfer adversarial network. Initially, the 1-D convolutional neural network is employed as the basic architecture. By training the domain discriminator and feature generator with an adversarial strategy, the recognition ability of the domain discriminant network and the feature extraction ability of the feature generation network can be enhanced. After that, a weighted learning strategy is introduced to guide the model to learn the cross-domain invariant feature. Also, a specific target domain classifier is utilized to redivide the target domain decision boundary to accurately classify the unlabeled target domain samples. Finally, five mainstream deep neural network methods are taken for comparison using the data from Western Reserve University and the motor-magnetic brake test designed by us. The results show that the proposed method reaches 90.18% and 96.53% classification accuracy on two datasets, respectively, which demonstrates superior performance compared with the state-of-the-art methods.

YOLO-STOD: an industrial conveyor belt tear detection model based on Yolov5 algorithm

Real-time detection of conveyor belt tearing is of great significance to ensure mining in the coal industry. The longitudinal tear damage problem of conveyor belts has the characteristics of multi-scale, abundant small targets, and complex interference sources. Therefore, in order to improve the performance of small-size tear damage detection algorithms under complex interference, a visual detection method YOLO-STOD based on deep learning was proposed. Firstly, a multi-case conveyor belt tear dataset is developed for complex interference and small-size detection. Second, the detection method YOLO-STOD is designed, which utilizes the BotNet attention mechanism to extract multi-dimensional tearing features, enhancing the model’s feature extraction ability for small targets and enables the model to converge quickly under the conditions of few samples. Secondly, Shape_IOU is utilized to calculate the training loss, and the shape regression loss of the bounding box itself is considered to enhance the robustness of the model. The experimental results fully proved the effectiveness of the YOLO-STOD detection method, which constantly surpasses the competing methods and achieves 91.2%, 91.9%, and 190.966 detection accuracy and detection speed in terms of recall, Map value, and FPS, respectively, which is able to satisfy the needs of industrial real-time detection and is expected to be used in the real-time detection of conveyor belt tearing in the industrial field.

Lightweight Multi-Scale Network for Segmentation of Riverbank Sand Mining Area in Satellite Images

Riverbank sand overexploitation is threatening the ecology and shipping safety of rivers. The rapid identification of riverbank sand mining areas from satellite images is extremely important for ecological protection and shipping management. Image segmentation methods based on AI technology are gradually becoming popular in academia and industry. However, traditional neural networks have complex structures and numerous parameters, making them unsuitable for meeting the needs of rapid extraction in large areas. To improve efficiency, we proposed a lightweight multi-scale network (LMS Net), which uses a lightweight multi-scale (LMS) block in both the encoder and decoder. The lightweight multi-scale block combines parallel computing and depthwise convolution to reduce the parameters of the network and enhance its multi-scale extraction ability. We created a benchmark dataset to validate the accuracy and efficiency improvements of our network. Comparative experiments and ablation studies proved that our LMS Net is more efficient than traditional methods like Unet and more accurate than typical lightweight methods like Ghostnet and other more recent methods. The performance of our proposed network meets the requirements of river management.

YOLOv7-DWS: tea bud recognition and detection network in multi-density environment via improved YOLOv7

IntroductionAccurate detection and recognition of tea bud images can drive advances in intelligent harvesting machinery for tea gardens and technology for tea bud pests and diseases. In order to realize the recognition and grading of tea buds in a complex multi-density tea garden environment.MethodsThis paper proposes an improved YOLOv7 object detection algorithm, called YOLOv7-DWS, which focuses on improving the accuracy of tea recognition. First, we make a series of improvements to the YOLOv7 algorithm, including decouple head to replace the head of YOLOv7, to enhance the feature extraction ability of the model and optimize the class decision logic. The problem of simultaneous detection and classification of one-bud-one-leaf and one-bud-two-leaves of tea was solved. Secondly, a new loss function WiseIoU is proposed for the loss function in YOLOv7, which improves the accuracy of the model. Finally, we evaluate different attention mechanisms to enhance the model’s focus on key features.Results and discussionThe experimental results show that the improved YOLOv7 algorithm has significantly improved over the original algorithm in all evaluation indexes, especially in the RTea(+6.2%) and mAP@0.5 (+7.7%). From the results, the algorithm in this paper helps to provide a new perspective and possibility for the field of tea image recognition.

Deep Learning-Based Pointer Meter Reading Recognition for Advancing Manufacturing Digital Transformation Research.

With the digital transformation of the manufacturing industry, data monitoring and collecting in the manufacturing process become essential. Pointer meter reading recognition (PMRR) is a key element in data monitoring throughout the manufacturing process. However, existing PMRR methods have low accuracy and insufficient robustness due to issues such as blur, uneven illumination, tilt, and complex backgrounds in meter images. To address these challenges, we propose an end-to-end PMRR method based on a decoupled circle head detection algorithm (YOLOX-DC) and a Unet-like pure Transformer segmentation network (PM-SwinUnet). First, according to the characteristics of the pointer dial, the YOLOX-DC detection algorithm is designed based on the exceeding you only look once detector (YOLOX). The decoupled circle head of YOLOX-DC detects the pointer meter dial more accurately than the commonly used rectangular detection head. Second, the window multi-head attention of the PM-SwinUnet network enhances the feature extraction ability of pointer meter images and solves problems of missed scale detection and incomplete pointer segmentation. Additionally, the scale and pointer fitting module is introduced into the PM-SwinUnet to locate the accurate position of the scale and pointer. Finally, through the angle relationship between the pointer and the first two main scale lines, the pointer meter reading is accurately calculated by the improved angle method. Experimental results demonstrate the effectiveness and superiority of the proposed end-to-end method across three-pointer meter datasets. Furthermore, it provides a rapid and robust approach to the digital transformation of manufacturing systems.

Interplay between conformational flexibility, intermolecular H-bonding and 3d-metal cation extraction ability in a series of thiacalix[4]arene lower rim disubstituted Schiff base derivatives.

The rational design of organic ligands with the aim to control their binding abilities towards different metal ions can be considered as one of the key concepts in supramolecular coordination chemistry. Regarding the macrocyclic compounds of thiacalix[4]arene family, this can be achieved via the targeted modulation of macrocyclic platform rigidity as well as the proper choice of appended binding sites. Four macrocyclic salen-type ligands based on lower rim disubstituted thiacalix[4]arene derivatives, adopted in a cone conformation, bearing highly coordinating iminophenolic or catecholic groups and two -CH2- moieties as spacers but presenting different abilities to form H-bonds, were chosen to elucidate the interplay between the conformational flexibility of the macrocyclic ligands, propensity to participate in the intermolecular H-bonding and the extraction ability of 3d-metal cations. X-ray diffraction analysis, theoretical DFT calculations, IR and Raman spectroscopies, and dynamic light scattering (DLS) studies performed in combination with liquid-liquid metal extraction study revealed that compounds 4, and 6, based on a thiacalix[4]arene macrocyclic platform, display a higher extraction ability towards all studied 3d-metal ions, caused by enhanced conformational flexibility. This is in good accordance with the ability of 6 to form H-bonded supramolecular assemblies in solution and crystalline phases due to recognition between the catecholic moieties.

Rolling bearing fault diagnosis method based on gramian angular difference field and dynamic self-calibrated convolution module.

To address the problem of insufficient feature extraction abilities of traditional fault diagnosis methods under conditions of sample scarcity and strong noise interference, a rolling bearing fault diagnosis method based on the Gramian Angular Difference Field (GADF) and Dynamic Self-Calibrated Convolution (DSC) is proposed. First, the GADF method converts one-dimensional signals into GADF images to capture nonlinear relationships and periodic information in time-series data. Second, a dynamic self-calibrated convolution module is introduced to enhance the feature extraction ability of the model. The DSC module dynamically adjusts the weights of parallel convolution kernels based on real-time data characteristics, effectively improving the feature extraction ability and generalization performance of the model. Finally, the proposed method is validated using bearing datasets from Huazhong University of Science and Technology and Harbin Institute of Technology, and is compared with other advanced models. The results show that the classification accuracy of the proposed method is basically above 90% when adding Gaussian white noise with a signal-to-noise ratio of -8 dB, which is a significant improvement of 6%-15% compared with other models. Therefore, the proposed method has excellent diagnostic performance in the rolling bearing fault diagnosis task under strong noise and small training samples.

Pixel-level Identification and Collapse of Post-earthquake Building Complexes Based on Satellite Remote Sensing Images

Due to the obvious diversity and complexity of damage patterns, geometries, and spatial scales of urban building complex earthquake hazards, conventional identification and assessment methods are less generalizable in real post-earthquake scenarios. Compared with time-range signals such as kinetic acceleration, image/video data provide a new source of perceptual information for accurately assessing the post-earthquake damage of urban building complexes. In order to realize the integrated, comprehensive, and rapid identification and assessment of the structural damage of urban buildings after an earthquake, this paper proposes a geometrically constrained deep learning framework for seismic damage identification and assessment of buildings based on computer vision, and systematically carries out a multi-scale seismic damage identification and assessment method that associates the “building group, building unit, and structural component" with the "building group, building unit, and structural component”. This paper proposes a geometrically constrained deep learning framework for seismic damage identification and assessment based on computer vision, and systematically carries out the research on the identification and assessment of “building groups-building units-structural components”. A method for fine identification of densely distributed small-target buildings and rapid assessment of collapse state after earthquake based on satellite remote sensing images at high altitude is proposed. A semantic segmentation network for post-earthquake building cluster identification and assessment is built, the influence law of the weight coefficients of the GCE loss on the segmentation performance of the model is systematically investigated, and the geometric feature optimization performance of the GCE loss in the training process and the multi-level feature extraction ability are analyzed, which verifies the effectiveness and accuracy of the geometrically constrained deep learning method for multi-scale seismic damage identification of buildings proposed in this paper.

A Grounding Current Prediction Method Based on Frequency-Enhanced Transformer

Concerning the problem that the coupling relationship in substation scenarios is complex and the Transformer model makes it difficult to capture the correlation between multiple variables of grounding current, resulting in low accuracy of grounding current prediction, a ground current prediction method based on frequency-enhanced Transformer is proposed. Firstly, in the data preprocessing stage, the best frequency domain decomposition algorithm is designed to obtain the high-frequency and low-frequency component data containing different component features so as to enhance the initial features that the model focuses on. Secondly, the data slicing and embedding module is designed to replace the original embedding module of the Transformer to realize the enhanced extraction of local features of the data. Finally, in the feature extraction stage, an enhanced attention mechanism is introduced to replace the standard attention mechanism to capture the intrinsic features of the sequence time dimension and the variable dimension in parallel so as to improve the extraction ability of Transformer multivariate features. Experimental results on the self-built grounding current dataset and the public dataset show that the proposed method outperforms existing advanced methods, verifying the effectiveness of the proposed method.

Pear Fruit Detection Model in Natural Environment Based on Lightweight Transformer Architecture

Aiming at the problems of low precision, slow speed and difficult detection of small target pear fruit in a real environment, this paper designs a pear fruit detection model in a natural environment based on a lightweight Transformer architecture based on the RT-DETR model. Meanwhile, Xinli No. 7 fruit data set with different environmental conditions is established. First, based on the original model, the backbone was replaced with a lightweight FasterNet network. Secondly, HiLo, an improved and efficient attention mechanism with high and low-frequency information extraction, was used to make the model lightweight and improve the feature extraction ability of Xinli No. 7 in complex environments. The CCFM module is reconstructed based on the Slim-Neck method, and the loss function of the original model is replaced with the Shape-NWD small target detection mechanism loss function to enhance the feature extraction capability of the network. The comparison test between RT-DETR and YOLOv5m, YOLOv7, YOLOv8m and YOLOv10m, Deformable-DETR models shows that RT-DETR can achieve a good balance in terms of model lightweight and recognition accuracy compared with other models, and comprehensively exceed the detection accuracy of the current advanced YOLOv10 algorithm, which can realize the rapid detection of Xinli No. 7 fruit. In this paper, the accuracy rate, recall rate and average accuracy of the improved model reached 93.7%, 91.9% and 98%, respectively, and compared with the original model, the number of params, calculation amount and weight memory was reduced by 48.47%, 56.2% and 48.31%, respectively. This model provides technical support for Xinli No. 7 fruit detection and model deployment in complex environments.

Investigation into the Prediction of Ship Heave Motion in Complex Sea Conditions Utilizing Hybrid Neural Networks

While navigating at sea, ships are influenced by various factors, including wind, waves, and currents, which can result in heave motion that significantly impacts operations and potentially leads to accidents. Accurate forecasting of ship heaving is essential to guarantee the safety of maritime navigation. Consequently, this paper proposes a hybrid neural network method that combines Convolutional Neural Networks (CNNs), Bidirectional Long Short-Term Memory Networks (BiLSTMs), and an Attention Mechanism to predict the heaving motion of ships in moderate to complex sea conditions. The data feature extraction ability of CNNs, the temporal analysis capabilities of BiLSTMs, and the dynamic adjustment function of Attention on feature weights were comprehensively utilized to predict a ship’s heave motion. Simulations of a standard container ship’s motion time series under complex sea state conditions were carried out. The model training and validation results indicate that, under sea conditions 4, 5, and 6, the CNN-BiLSTM-Attention method demonstrated significant improvements in MAPE, APE, and RMSE when compared to the traditional LSTM, Attention, and LSTM-Attention methods. The CNN-BiLSTM-Attention method could enhance the accuracy of the prediction. Heave displacement, pitch displacement, pitch velocity, pitch acceleration, and incoming wave height were chosen as key input features. Sensitivity analysis was conducted to optimize the prediction performance of the CNN-BiLSTM-Attention hybrid neural network method, resulting in a significant improvement in MAPE and enhancing the accuracy of ship motion prediction. The research presented in this paper establishes a foundation for future studies on ship motion prediction.

FCAE-YOLOv8n: a target detection method for immature grape clusters

ABSTRACT In order to quickly and accurately detect immature grape clusters in the complex environment of orchard, a new model FCAE-YOLOv8n was proposed with low hardware requirements, high target accuracy and fast model convergence. The model was improved based on the original YOLOv8n model. Firstly, the model possessed numerous parameters and was unsuitable for mobile terminal deployment. Faster Block module was used to replace Bottleneck module in C2f, which reduced model parameters. Secondly, complex similarly-colored backgrounds, numerous small-sized grape grains, and mutual occlusion among grape clusters contributed to its poor recognition. The CA mechanism was embedded in the model to improve its feature extraction ability. Finally, the EIoU loss function instead the CIoU loss function was used to accelerate the convergence of the bounding interval. FCAE-YOLOv8n was experimentally compared with different models using the self-build datasets. FCAE-YOLOv8n model achieved precision of 98.6%, recall rate of 97.1% and mAP@0.5:0.95 of 92.5%. It increases by 1.2%, 2.0%, and 3.4% respectively compared with the original YOLOv8n. The average detection speed of each image was increased. It achieved the rapid and precise recognition of immature grape clusters in complex orchards, which provided technical support for subsequent automated bagging.

Fraction and Antibacterial Activity of Bitter Melons (Momordica charantia) in Koi Fish (Cypinus carpio) Infected with Aeromonas salmonicida Bacteria and its Effect on Gills Histopathology

Koi fish (Cyprinus carpio) is a type of freshwater ornamental fish with significant economic value, attracting widespread public interest due to its vibrant colors and agile movements. Therefore, this study introduces a novel approach by evaluating the antibacterial activity of bitter melon extract fraction (Momordica charantia) against A. salmonicida infection in koi fish and its impact on gill histopathology. The novelty of this study lies in the exploration of bitter melon extract fraction as a natural, eco-friendly antibacterial alternative to synthetic antibiotics in aquaculture. This research is crucial as it addresses the growing global concern over antibiotic resistance and the urgent need for sustainable, natural solutions in fish health management. Antibacterial tests were conducted to assess the ability of the bitter melon extract to inhibit bacterial growth, while histopathological analysis aimed to observe the specific changes in gill tissues of infected koi fish. The results revealed that the bitter melon extract fraction significantly reduced bacterial count and improved gill tissue integrity. Histopathological improvements were observed, including reduced hyperplasia, lamellar fusion, edema, and necrosis, indicating tissue recovery. This finding highlights the potential of bitter melon extract fraction as a natural antibacterial agent for controlling bacterial infections in fish, offering an alternative to chemical antibiotics. Further research is needed to determine the optimal dosage and to explore its antibacterial mechanism, ensuring its safe and effective application in aquaculture.

Adaptive cascade decoders for segmenting challenging regions in medical images.

CNN-based techniques have achieved impressive outcomes in medical image segmentation but struggle to capture long-term dependencies between pixels. The Transformer, with its strong feature extraction and representation learning abilities, performs exceptionally well within the domain of medical image partitioning. However, there are still shortcomings in bridging local to global connections, resulting in occasional loss of positional information. To address this, we introduce a decoder based on dynamic convolution, called Adaptive Cascade Decoder (ACD). It can adaptively adjust the receptive field size based on medical images, adapting a set of parameters for each medical image individually. The ACD consists of an Adaptive Attention (ADA) and a Multi-Scale Convolution module (MSC). By enhancing feature extraction from local to global scales, it addresses issues of diminished contrast and fuzzy boundaries common in medical image segmentation. While increasing contextual connections, it also reduces certain parameters, thereby lowering memory consumption. Our model, T-ACD, uses the encoder backbone of TransUNet, which chunks feature maps from convolutional neural networks and feeds them as one-dimensional sequences into the Transformer. This leverages the Transformer's prowess in handling sequences, further refining the extracted features. In experiments involving heart and multi-organ segmentation, T-ACD excels in segmenting challenging areas. On the ACDC dataset, we achieve a DICE coefficient of 92.02%. For the most challenging right ventricle, it is improved to 90.68%, which is increased by 5.17%. In the realm of medical segmentation, the core design of ACD can be generalized to other challenging organ segmentations.

Graph pooling for graph-level representation learning: a survey

In graph-level representation learning tasks, graph neural networks have received much attention for their powerful feature learning capabilities. However, with the increasing scales of graph data, how to efficiently process and extract the key information has become the focus of research. The graph pooling technique, as a key step in graph neural networks, simplifies the graph structure by merging nodes or subgraphs, which significantly improves the computational efficiency and feature extraction ability of graph neural networks. Although various graph pooling methods have been proposed by numerous scholars, there is still a relative lack of systematic summaries of these works. In this paper, we comprehensively sort out the fundamentals and recent progress of graph pooling techniques in graph neural networks and discuss its wide range of application scenarios, as well as the current challenges and opportunities, which point out the direction for future research. Specifically, we first provide a detailed introduction to the basics of graph pooling, including its definition, principles, and its function in graph neural networks. Then, we categorize and summarize the research preliminaries of graph pooling, including various graph pooling methods proposed in recent years. Next, we explore the potential of graph pooling for a wide range of applications, which provides insightful insights for the promotion and practice of graph pooling technology in more fields. Furthermore, we conduct a comparative analysis of various graph pooling methods and evaluate their performance on a benchmark dataset, providing a comprehensive understanding of their strengths and weaknesses. Finally, we systematically analyze the challenges and opportunities of the current graph pooling methods and provide a prospective outlook on future research directions.

Fault diagnosis method of rolling bearing under variable operating conditions based on MFCCNN

A rolling bearing fault diagnosis method based on multi-scale feature fusion and cross-level connection convolutional neural network (MFCCNN) is proposed to address the issues of low accuracy and low generalization performance caused by the complex and variable actual working conditions of rolling bearings. Firstly, a Markov transition field (MTF) is used to convert a one-dimensional signal into a two-dimensional image with temporal correlation. Next, an improved channel attention mechanism is integrated into the multi-scale network to assign different weights to each channel information and extract initial features. Then, a position attention mechanism is created and incorporated into the cross-level connection to enhance the feature expression ability for deep feature extraction. The MFCCNN faults diagnosis model is developed based on the above module. Finally, the MTF images is input into the proposed model for training, and the Softmax classifier is utilized to achieve fault classification. To verify the effectiveness and superiority of the MFCCNN method, the CWRU and MFS rolling bearing dataset are selected for experimental verification. The results demonstrates that the MFCCNN model has higher diagnostic accuracy, stronger noise immunity, and generalization performance in varying operating conditions compared to other fault diagnosis methods.

Adaptive Ramanujan package decomposition method: a novel precise health indicator of rolling bearings

ABSTRACT In this paper, an adaptive Ramanujan packet decomposition (ARPD) method based on precise envelope segmentation is proposed to improve the accuracy of weak fault feature extraction and realise early non-destructive testing and condition evaluation of rolling bearings. Firstly, the power spectral density of the original signal is envelope-processed by accurate envelope segmentation method, so as to reduce the extreme points produced by interference frequency, and then realise the optimal frequency band division and reduce the influence of noise components. Secondly, the Ramanujan spectrum signal-to-noise ratio (RSSNR) index is introduced to quantify the fault characteristic information of the corresponding component in each frequency band. The largest component of the index is selected as the optimal mode, and the current number of layers is defined as the optimal number of modes. Finally, through the idea of iterative decomposition and RSSNR index, the problem that the optimal mode number is difficult to determine accurately is eliminated, and the optimal mode number is determined adaptively. The results of simulation and experimental signal analysis show that ARPD method can accurately divide the frequency bands and has excellent weak fault feature extraction ability, so it is an effective early non-destructive testing and condition evaluation method for rolling bearings.

D3-YOLOv10: Improved YOLOv10-BasedLightweight Tomato Detection Algorithm Under Facility Scenario

Accurate and efficient tomato detection is one of the key techniques for intelligent automatic picking in the area of precision agriculture. However, under the facility scenario, existing detection algorithms still have challenging problems such as weak feature extraction ability for occlusion conditions and different fruit sizes, low accuracy on edge location, and heavy model parameters. To address these problems, this paper proposed D3-YOLOv10, a lightweight YOLOv10-based detection framework. Initially, a compact dynamic faster network (DyFasterNet) was developed, where multiple adaptive convolution kernels are aggregated to extract local effective features for fruit size adaption. Additionally, the deformable large kernel attention mechanism (D-LKA) was designed for the terminal phase of the neck network by adaptively adjusting the receptive field to focus on irregular tomato deformations and occlusions. Then, to further improve detection boundary accuracy and convergence, a dynamic FM-WIoU regression loss with a scaling factor was proposed. Finally, a knowledge distillation scheme using semantic frequency prompts was developed to optimize the model for lightweight deployment in practical applications. We evaluated the proposed framework using a self-made tomato dataset and designed a two-stage category balancing method based on diffusion models to address the sample class-imbalanced issue. The experimental results demonstrated that the D3-YOLOv10 model achieved an mAP0.5 of 91.8%, with a substantial reduction of 54.0% in parameters and 64.9% in FLOPs, compared to the benchmark model. Meanwhile, the detection speed of 80.1 FPS more effectively meets the demand for real-time tomato detection. This study can effectively contribute to the advancement of smart agriculture research on the detection of fruit targets.