Ability Of Model Research Articles

Automatic Apple Detection and Counting with AD-YOLO and MR-SORT.

In the production management of agriculture, accurate fruit counting plays a vital role in the orchard yield estimation and appropriate production decisions. Although recent tracking-by-detection algorithms have emerged as a promising fruit-counting method, they still cannot completely avoid fruit occlusion and light variations in complex orchard environments, and it is difficult to realize automatic and accurate apple counting. In this paper, a video-based multiple-object tracking method, MR-SORT (Multiple Rematching SORT), is proposed based on the improved YOLOv8 and BoT-SORT. First, we propose the AD-YOLO model, which aims to reduce the number of incorrect detections during object tracking. In the YOLOv8s backbone network, an Omni-dimensional Dynamic Convolution (ODConv) module is used to extract local feature information and enhance the model's ability better; a Global Attention Mechanism (GAM) is introduced to improve the detection ability of a foreground object (apple) in the whole image; a Soft Spatial Pyramid Pooling Layer (SSPPL) is designed to reduce the feature information dispersion and increase the sensory field of the network. Then, the improved BoT-SORT algorithm is proposed by fusing the verification mechanism, SURF feature descriptors, and the Vector of Local Aggregate Descriptors (VLAD) algorithm, which can match apples more accurately in adjacent video frames and reduce the probability of ID switching in the tracking process. The results show that the mAP metrics of the proposed AD-YOLO model are 3.1% higher than those of the YOLOv8 model, reaching 96.4%. The improved tracking algorithm has 297 fewer ID switches, which is 35.6% less than the original algorithm. The multiple-object tracking accuracy of the improved algorithm reached 85.6%, and the average counting error was reduced to 0.07. The coefficient of determination R2 between the ground truth and the predicted value reached 0.98. The above metrics show that our method can give more accurate counting results for apples and even other types of fruit.

High-Precision Detection of Lung Adenocarcinoma Using Augmented VGG16 and Transfer Learning

Objectives: The research aims to enhance the effectiveness of the deep learning model for detecting lung adenocarcinoma using VGG16 CNN with transfer learning on medical images. The proposed model addresses the overfitting issues of existing models, thereby improving lung cancer detection accuracy. Methods: The lung CT scan images are analyzed for detection and classification using VGG16, leveraging transfer learning techniques. The dataset includes labeled lung images from LUNA 16, Kaggle, and other datasets, which are augmented to introduce variability and reduce overfitting. The images are resized format to 224 × 224 pixels. During the validation and training phases, the model's performance is evaluated based on accuracy, and precision. Findings: The adjusted VGG16 model attains a training accuracy of 99.42%, a validation accuracy of 99.13%, and a validation precision of 95.45%, indicating strong generalization capabilities, outperforming other existing models. Novelty: This research presents a new method for detecting lung cancer by combining advanced preprocessing and data augmentation techniques with the VGG16 architecture using transfer learning. This combination greatly enhances the model's ability to detect cancer accurately, setting a new standard in the field. Keywords: Lung cancer, VGG16, Transfer learning, Data augmentation, medical imaging, CT (Computer Tomography)

A Novel Triad of Bio-Inspired Design, Digital Fabrication, and Bio-Derived Materials for Personalised Bone Repair.

The emerging paradigm of personalised bone repair embodies a transformative triad comprising bio-inspired design, digital fabrication, and the exploration of innovative materials. The increasing average age of the population, alongside the rising incidence of fractures associated with age-related conditions such as osteoporosis, necessitates the development of customised, efficient, and minimally invasive treatment modalities as alternatives to conventional methods (e.g., autografts, allografts, Ilizarov distraction, and bone fixators) typically employed to promote bone regeneration. A promising innovative technique involves the use of cellularised scaffolds incorporating mesenchymal stem cells (MSCs). The selection of materials-ranging from metals and ceramics to synthetic or natural bio-derived polymers-combined with a design inspired by natural sources (including bone, corals, algae, shells, silk, and plants) facilitates the replication of geometries, architectures, porosities, biodegradation capabilities, and mechanical properties conducive to physiological bone regeneration. To mimic internal structures and geometries for construct customisation, scaffolds can be designed using Computer-aided Design (CAD) and fabricated via 3D-printing techniques. This approach not only enables precise control over external shapes and internal architectures but also accommodates the use of diverse materials that improve biological performance and provide economic advantages. Finally, advanced numerical models are employed to simulate, analyse, and optimise the complex processes involved in personalised bone regeneration, with computational predictions validated against experimental data and in vivo studies to ascertain the model's ability to predict the recovery of bone shape and function.

Risk assessment of urban infrastructure vulnerability to meteorological disasters: A case study of Dongguan, China

Effective forecasting and response to meteorological hazards are crucial for safeguarding life, property, and supporting sustainable socioeconomic development. With the rising frequency and severity of meteorological hazards worldwide, this study proposes an enhanced risk assessment framework for urban infrastructure exposed to extreme weather events, with a focus on cascading impacts to critical services such as electricity, communication, and transportation networks (roads and subways). A disaster-loss model is developed to quantify infrastructure vulnerability at various spatial and temporal scales under heavy rainfall conditions, accounting for secondary effects. The model's performance is validated through empirical analysis of a 15-year rainfall event in Dongguan City, China, occurring from September 7–8, 2023. Results indicate the model's ability to predict real-event outcomes with approximately 70% accuracy. This model offers valuable insights for disaster prevention and mitigation strategies, aiding decision-makers in optimizing emergency resource allocation, enhancing disaster response efficiency, and issuing timely public risk warnings to minimize losses.

Financial Fraud Detection Study - Based on Logit Model

The problem of financial fraud has always been a key issue of concern for researchers at home and abroad, and financial fraud of listed companies occurs from time to time at home and abroad, which produces a huge loss of interest for investors as well as the downturn and instability of the capital market, for this reason, it is necessary to carry out a study on the detection of financial fraud. This paper analyzes the financial fraud detection of Chinese listed companies by establishing a Logit model, firstly, obtain the financial statement sample data of Chinese listed companies from CSMAR, and divide the sample into training set and test set with the ratio of 9:1 to preprocess the sample data and fill in the missing values; secondly, this paper selects the financial statement data, and based on the previous research, selects the characteristics that are related to the risk of financial fraud and constitute the risk of financial fraud. high features, and constitute the feature indicators for financial fraud detection; again, since detecting financial fraud is a binary classification problem, the sample is divided into fraudulent companies and normal companies, so this paper studies the financial fraud problem through the discrete choice model in the econometric model, constructs a Logit model, conducts a goodness-of-fit test on the sample data of CSMAR, and estimates the parameters using the method of maximum likelihood estimation ; finally, the test set data is used to test the model's ability to predict financial fraud.

Multi‐scale feature extraction for energy‐efficient object detection in remote sensing images

AbstractObject detection in remote sensing images aims to interpret images to obtain information on the category and location of potential targets, which is of great importance in traffic detection, marine supervision, and space reconnaissance. However, the complex backgrounds and large scale variations in remote sensing images present significant challenges. Traditional methods relied mainly on image filtering or feature descriptor methods to extract features, resulting in underperformance. Deep learning methods, especially one‐stage detectors, for example, the Real‐Time Object Detector (RTMDet) offers advanced solutions with efficient network architectures. Nevertheless, difficulty in feature extraction from complex backgrounds and target localisation in scale variations images limits detection accuracy. In this paper, an improved detector based on RTMDet, called the Multi‐Scale Feature Extraction‐assist RTMDet (MRTMDet), is proposed which address limitations through enhancement feature extraction and fusion networks. At the core of MRTMDet is a new backbone network MobileViT++ and a feature fusion network SFC‐FPN, which enhances the model's ability to capture global and multi‐scale features by carefully designing a hybrid feature processing unit of CNN and a transformer based on vision transformer (ViT) and poly‐scale convolution (PSConv), respectively. The experiment in DIOR‐R demonstrated that MRTMDet achieves competitive performance of 62.2% mAP, balancing precision with a lightweight design.

Personalized blood pressure estimation using multiview fusion information of wearable physiological signals and transfer learning

Continuous blood pressure (BP) monitoring is crucial for individual health management, yet the significant inter-individual variations among patients pose challenges to achieving precision medicine. In response to this issue, we propose a parallel cross-hybrid architecture that integrates a convolutional neural network backbone and a Mix-Transformer backbone. This model, grounded in multi-view physiological signals and personalized fine-tuning strategies, aims to estimate BP, facilitating the capture of physiological information across diverse receptive fields and enhancing network expressive capabilities. Our proposed architecture exhibits superior performance in estimating systolic blood pressure and diastolic blood pressure, with average absolute errors of 3.94 mmHg and 2.24 mmHg, respectively. These results surpass existing baseline models and align with the standards set by the British Hypertension Society, the Association for the Advancement of Medical Instrumentation, and the Institute of Electrical and Electronics Engineers for BP measurement. Additionally, this study explores a personalized model fine-tuning strategy by adjusting specific layers and incorporating individual information, presenting an optimal solution. The model's generalization ability is validated through transfer learning across databases (public and self-made). To enhance the proposed architecture's usability in wearable devices, this study employs a knowledge distillation strategy for model lightweighting, with preliminary application in our designed real-time BP estimation system. This study provides an efficient and accurate solution for personalized BP estimation, exhibiting broad potential applications.

Design of intelligent algorithm for object search based on IoT digital images

With the development of artificial intelligence, traditional object search and image recognition have been replaced by the Internet of Things and artificial intelligence. However, traditional object search algorithms often lack accuracy and low precision. Therefore, this study proposes a new intelligent encryption algorithm to address the issues of insufficient accuracy in object search algorithms and image recognition algorithms. The new algorithm ensures the security of user data and the response efficiency of the model during the conversation process by integrating fully homomorphic encryption technology and dynamic sparse attention mechanism. The dynamic sparse attention mechanism introduced simultaneously improves the model's ability to handle long sequence data by dynamically adjusting attention weights. Experimental results showed that the precision of the proposed algorithm was 0.05 % higher than that of random algorithms and 0.19 % higher than that of sorting algorithms. The recall rate of the proposed algorithm was 0.14 % higher than that of random algorithms and 0.16 % higher than that of sorting algorithms. The research algorithm can identify objects with certain characteristics and is suitable for specific environments, greatly reducing the probability of data leakage in object search and providing new ideas for research in this field.

MRS‐Net: Brain tumour segmentation network based on feature fusion and attention mechanism

AbstractAccurate segmentation of brain tumor magnetic resonance imaging (MRI) is crucial for treatment planning. Addressing the challenges of complex tumor structures and inadequate cross‐channel information utilization in Unet‐based segmentation, this paper proposes the multi‐scale residual brain tumor MRI segmentation network (MRS‐Net) incorporating an attention mechanism to enhance segmentation accuracy. First, the double residual feature fusion module is utilized to enhance the fusion of feature information between different levels. Second, the Atrous Spatial Pyramid Pooling is introduced as a bridging module of the network to capture the features at different scales of the image, so as to enhance the extraction capability of the network for detailed features. Finally, the inverted residual coordinate attention module replaces the direct splicing in Unet to fuse the large feature information at each level and scale, thus enhancing the model's ability to recognize the spatial location information of brain tumors. The Dice coefficients, positive predictive values (PPVs), sensitivities (Sensitivity) and Hausdorff distance (HD), which are the four evaluation indexes, reach 84.54%, 87.43%, 88.37% and 2.248, respectively, which are improved by 1.85%, 2.11%, 2.88% and 6.0%, respectively, compared with Unet. The experimental results show that MRS‐Net achieves better brain tumor image segmentation.

An inherently interpretable deep learning model for local explanations using visual concepts.

Over the past decade, deep learning has become the leading approach for various computer vision tasks and decision support systems. However, the opaque nature of deep learning models raises significant concerns about their fairness, reliability, and the underlying inferences they make. Many existing methods attempt to approximate the relationship between low-level input features and outcomes. However, humans tend to understand and reason based on high-level concepts rather than low-level input features. To bridge this gap, several concept-based interpretable methods have been developed. Most of these methods compute the importance of each discovered concept for a specific class. However, they often fail to provide local explanations. Additionally, these approaches typically rely on labeled concepts or learn directly from datasets, leading to the extraction of irrelevant concepts. They also tend to overlook the potential of these concepts to interpret model predictions effectively. This research proposes a two-stream model called the Cross-Attentional Fast/Slow Thinking Network (CA-SoftNet) to address these issues. The model is inspired by dual-process theory and integrates two key components: a shallow convolutional neural network (sCNN) as System-I for rapid, implicit pattern recognition and a cross-attentional concept memory network as System-II for transparent, controllable, and logical reasoning. Our evaluation across diverse datasets demonstrates the model's competitive accuracy, achieving 85.6%, 83.7%, 93.6%, and 90.3% on CUB 200-2011, Stanford Cars, ISIC 2016, and ISIC 2017, respectively. This performance outperforms existing interpretable models and is comparable to non-interpretable counterparts. Furthermore, our novel concept extraction method facilitates identifying and selecting salient concepts. These concepts are then used to generate concept-based local explanations that align with human thinking. Additionally, the model's ability to share similar concepts across distinct classes, such as in fine-grained classification, enhances its scalability for large datasets. This feature also induces human-like cognition and reasoning within the proposed framework.

Biomass prediction of sea lamprey population based on dynamic feature matrix composite differential equation system

This article presents a mathematical model to predict sea lamprey biomass, addressing the species' impact on the Great Lakes fishery. The model, comprising a Dynamic Age Pyramid and a Multi-species Lotka-Volterra system, captures the complex life stages and behaviors of sea lampreys. It simulates their diet, habits, and growth parameters across different ages, considering their adaptive sex ratio variation. Key findings include the model's ability to dynamically update sea lamprey population characteristics annually in response to environmental changes. Simulations show that populations with dynamic sex ratios reach equilibrium faster, with an increasing female proportion over time in resource-rich environments. However, these populations exhibit less resilience to rapid environmental shifts compared to those with stable sex ratios. The study concludes that the model offers a valuable tool for understanding complex organism-environment interactions and enhances traditional ecological models by providing a dynamic, biologically meaningful population feature matrix. This innovation aids in managing invasive species like sea lampreys without harming native ecosystems.

Bearing defect detection based on the improved YOLOv5 algorithm.

In the field of bearing defect detection, Aiming at the problem of low efficiency in manual inspection and prone to missed detections in scenarios with small target defects, and overlapping targets, an improved YOLOv5-based object detection method is proposed. Firstly, in terms of feature extraction, the C3 modules in the original backbone of YOLOv5 are replaced with the finer-grained Res2Block modules to improve the model's feature extraction ability. Secondly, in terms of feature fusion, a Bidirectional Feature Pyramid Network (BiFPN) is added to the original neck of YOLOv5 to enhance the fusion ability of shallow graphic features and deep semantic features. Finally, the performance of the improved YOLOv5 algorithm is validated through ablation experiments and comparative experiments with other defect detection algorithms, including the Small_obj algorithm the existing method of adding a small target detection head for identifying small target defects. The experimental results demonstrate that the improved YOLOv5 algorithm exhibits high mAP and accuracy in bearing defect detection, enabling more precise identification the types of small target defects on bearings in complex scenarios with multiple coexisting defects and overlapping detection targets, thereby providing valuable reference for practical bearing defect detection.

Identifying heart failure dynamics using multi-point electrocardiograms and deep learning

Abstract Background Heart failure (HF) hospitalizations are related to poor survival outcomes, emphasizing the need for early detection and intervention. Recent studies have reported the potential of deep learning algorithms in identifying heart failure through electrocardiogram (ECG) with notable accuracy. However, evidence supporting their utility in the ongoing monitoring and follow-up of HF patients remains uncertain. Advanced diagnostic tools that can effectively track heart failure progression or improvement over time are needed. Purpose To address this challenge, we developed a deep learning model analyzing 12-lead ECG waveforms at two time points, aiming to detect HF deterioration or improvement, potentially revolutionizing HF management. Methods This study analyzed 32,045 ECGs from 6,892 unique adult patients (aged 18 years and older) visiting the Department of Cardiology at Kobe University Hospital. The participants were randomly assigned to training (4,284 patients), validation (1,034 patients), and test (1,034 patients) datasets. ElecTransformer was developed to classify HF status into deteriorated, improved, and no change classes based on ECG waveform signals at two different time points and to estimate brain natriuretic peptide (BNP) levels. Performance metrics, such as the area under the receiver operating characteristic curve (AUROC) and accuracy, were calculated, and attention mapping via gradient-weighted class activation mapping was utilized to interpret the model's decision-making ability. Results The patients had an average age of 65 years (±15.4 years) and BNP levels averaged 71.5 pg/mL (25.8 to 185.2 pg/mL). For HF status classification, ElecTransformer achieved an AUROC of 0.845 and an accuracy of 0.821, improving to an AUROC of 0.885 and an accuracy of 0.833 with baseline BNP integration. The ElecTransformer accurately estimated BNP levels from ECG data, evidenced by a correlation coefficient of 0.798. Averaged attention mapping showed that the QRS complex was the most important feature for determining model outputs with lesser effects seen for P waves and T waves variance Conclusions Harnessing a Transformer-based deep learning model to analyze multiple ECGs markedly improves the accuracy in detecting changes in heart failure status. This method offers a valuable non-invasive option for monitoring heart failure progression, potentially leading to better patient management and earlier intervention in heart failure treatment.Architecture and Performance of ModelsAveraged attention mapping

Breast Lesion Detection for Ultrasound Images Using MaskFormer.

This study evaluates the performance of the MaskFormer model for segmenting and classifying breast lesions using ultrasound images, addressing ultrasound's limitations. Ultrasound used for breast cancer detection faces challenges like low image contrast and difficulty in the detection of small or multiple lesions, further complicated by variability based on operator skill. Initial experiments with U-Net and other CNN-based models revealed constraints, such as early plateauing in model loss, which indicated suboptimal learning and performance. In contrast, MaskFormer demonstrated continuous improvement, achieving higher precision in breast lesion segmentation and significantly reducing both false positives and false negatives. Comparative analysis showed MaskFormer's superior performance, with the highest precision and recall rates for malignant lesions and an overall mean average precision (mAP) of 0.943. The model's ability to detect a diverse range of breast lesions, including those potentially missed by the human eye, especially by less experienced practitioners, underscores its potential. These findings suggest that integrating AI models like MaskFormer could greatly enhance ultrasound performance for breast cancer detection, providing reliable, operator-independent image analysis and potentially improving patient outcomes on a global scale.

Study of Five-Hundred-Meter Aperture Spherical Telescope Feed Cabin Time-Series Prediction Studies Based on Long Short-Term Memory-Self-Attention.

The Five-hundred-meter Aperture Spherical Telescope (FAST), as the world's most sensitive single-dish radio telescope, necessitates highly accurate positioning of its feed cabin to utilize its full observational potential. Traditional positioning methods that rely on GNSS and IMU, integrated with TS devices, but the GNSS and TS devices are vulnerable to other signal and environmental disruptions, which can significantly diminish position accuracy and even cause observation to stop. To address these challenges, this study introduces a novel time-series prediction model that integrates Long Short-Term Memory (LSTM) networks with a Self-Attention mechanism. This model can hold the precision of feed cabin positioning when the measure devices fail. Experimental results show that our LSTM-Self-Attention model achieves a Mean Absolute Error (MAE) of less than 10 mm and a Root Mean Square Error (RMSE) of approximately 12 mm, with the errors across different axes following a near-normal distribution. This performance meets the FAST measurement precision requirement of 15 mm, a standard derived from engineering practices where measurement accuracy is set at one-third of the control accuracy, which is around 48 mm (according to the accuracy form the official threshold analysis on the focus cabin of FAST). This result not only compensates for the shortcomings of traditional methods in consistently solving feed cabin positioning, but also demonstrates the model's ability to handle complex time-series data under specific conditions, such as sensor failures, thus providing a reliable tool for the stable operation of highly sensitive astronomical observations.

DMDLK-Net: A dynamic multi-scale feature fusion network with deformable large kernel for medical segmentation

Abstract. Skin disease image segmentation is a crucial component of computer-aided diagnosis, providing precise localization and delineation of lesions that enhance diagnostic accuracy and efficiency. Despite significant advancements in convolutional neural networks (CNNs), there remains substantial room for improvement in segmentation performance due to the diverse and complex nature of skin lesions. In this study, we propose DMDLK-Net, a dynamic multi-scale feature fusion network with deformable large kernels, specifically designed to address the challenges in skin disease segmentation. Our network incorporates a Dynamic Deformable Large Kernel (DDLK) module and a Dynamic Multi-Scale Feature Fusion (DMFF) module, enhancing the model's ability to capture intricate lesion features. We present the performance of DMDLK-Net on the ISIC-2018 dataset, highlighting its promising results. Key contributions of this work include the innovative use of deformable large kernels for adaptive feature extraction and the introduction of dynamic multi-scale fusion to balance local and global information. Our experimental results confirm the effectiveness of DMDLK-Net in delivering high-precision segmentation, thus providing a reliable tool for clinical applications.

Construction of a prediction model for induction of labor based on a small sample of clinical indicator data

Because of the diversity and complexity of clinical indicators, it is difficult to establish a comprehensive and reliable prediction model for induction of labor (IOL) outcomes with existing methods. This study aims to analyze the clinical indicators related to IOL and to develop and evaluate a prediction model based on a small-sample of data. The study population consisted of a total of 90 pregnant women who underwent IOL between February 2023 and January 2024 at the Shanghai First Maternity and Infant Healthcare Hospital, and a total of 52 clinical indicators were recorded. Maximal information coefficient (MIC) was used to select features for clinical indicators to reduce the risk of overfitting caused by high-dimensional features. Then, based on the features selected by MIC, the support vector machine (SVM) model based on small samples was compared and analyzed with the fully connected neural network (FCNN) model based on large samples in deep learning, and the receiver operating characteristic (ROC) curve was given. By calculating the MIC score, the final feature dimension was reduced from 55 to 15, and the area under curve (AUC) of the SVM model was improved from 0.872 before feature selection to 0.923. Model comparison results showed that SVM had better prediction performance than FCNN. This study demonstrates that SVM successfully predicted IOL outcomes, and the MIC feature selection effectively improves the model's generalization ability, making the prediction results more stable. This study provides a reliable method for predicting the outcome of induced labor with potential clinical applications.

Forecasting Wholesale Electricity Market Prices Considering Bidding Conditions Using Price Sensitivity

This article proposes a sophisticated forecasting model to predict significant price volatility on the Japan Electric Power Exchange (JEPX) due to the growing influence of solar photovoltaics (PV) in the energy mix. As solar power generation continues to grow, it significantly impacts the daily and seasonal price trends in the market. This method employs a neural network that integrates daily prices, net electricity demand (total demand minus PV output), and underlying time‐series data to determine the objective variable, defined as the deviation from the average price over the past week. Incorporating price sensitivity data from JEPX increases the accuracy of the model. These data reflect how prices respond to market‐specified bid adjustments, including a variety of bidding, providing a nuanced understanding of market responses. The results demonstrate that adding multiple price sensitivities significantly improves the model's ability to detect price spikes and provides a robust tool for transmission and distribution system operators to manage risk and optimize their market strategies in an increasingly renewable energy‐dominated landscape. This approach not only addresses daily and seasonal price fluctuations but also aligns with broader sustainability goals as the share of solar PV generation continues to grow.

Enhancing bond strength prediction at UHPC-NC interface: A data-driven approach with augmentation and explainability

Existing concrete structures often have difficulty reaching their design service life due to aging, increased loads, and natural disasters, and they need to be repaired and strengthened or replaced. Ultra-high-performance concrete (UHPC), known for its ultra-high strength and excellent toughness, offers a promising solution for repairing and strengthening normal concrete (NC) structures. It is expected to be a viable solution when applied to the repair and strengthening of NC structures. A reliable interfacial bonding between the two materials (NC substrate and UHPC) determines the overall performance of this composite structure. In this study, a data-driven approach is proposed to predict the bond performance at the UHPC-NC interface as accurately as possible. To overcome the problem of limited experimental data, a data augmentation model is introduced, combining kernel density estimation (KDE) and tabular generative adversarial networks (TGAN). The optimal model is determined from six decision tree-based ensemble learning models applied to two strategies: "synthetic training - real testing" and "real training - real testing." Finally, a parametric analysis is performed using SHapley Additive exPlanations (SHAP) to elucidate the importance and sensitivity of different features related to bond strength. The findings illustrate that the suggested KDE-TGAN model effectively captures the distribution of the original dataset, enhancing both the accuracy and robustness of the bond strength prediction models. Furthermore, the model's ability to explain the importance and sensitivity of different features provides valuable insights into bond strength prediction. Thus, the proposed data augmentation model provides a reliable approach for modeling experimental data with small samples in structural engineering applications.

CLIP-DFGS: A Hard Sample Mining Method for CLIP in Generalizable Person Re-Identification

Recent advancements in pre-trained vision-language models like CLIP have shown promise in person re-identification (ReID) applications. However, their performance in generalizable person re-identification tasks remains suboptimal. The large-scale and diverse image-text pairs used in CLIP's pre-training may lead to a lack or insufficiency of certain fine-grained features. In light of these challenges, we propose a hard sample mining method called DFGS (Depth-First Graph Sampler), based on depth-first search, designed to offer sufficiently challenging samples to enhance CLIP's ability to extract fine-grained features. DFGS can be applied to both the image encoder and the text encoder in CLIP. By leveraging the powerful cross-modal learning capabilities of CLIP, we aim to apply our DFGS method to extract challenging samples and form mini-batches with high discriminative difficulty, providing the image model with more efficient and challenging samples that are difficult to distinguish, thereby enhancing the model's ability to differentiate between individuals. Our results demonstrate significant improvements over other methods, confirming the effectiveness of DFGS in providing challenging samples that enhance CLIP's performance in generalizable person re-identification.