A CNN-CBAM-BIGRU model for protein function prediction.

Understanding a protein's function based solely on its amino acid sequence is a crucial but intricate task in bioinformatics. Traditionally, this challenge has proven difficult. However, recent years have witnessed the rise of deep learning as a powerful tool, achieving significant success in protein function prediction. Their strength lies in their ability to automatically learn informative features from protein sequences, which can then be used to predict the protein's function. This study builds upon these advancements by proposing a novel model: CNN-CBAM+BiGRU. It incorporates a Convolutional Block Attention Module (CBAM) alongside BiGRUs. CBAM acts as a spotlight, guiding the CNN to focus on the most informative parts of the protein data, leading to more accurate feature extraction. BiGRUs, a type of Recurrent Neural Network (RNN), excel at capturing long-range dependencies within the protein sequence, which are essential for accurate function prediction. The proposed model integrates the strengths of both CNN-CBAM and BiGRU. This study's findings, validated through experimentation, showcase the effectiveness of this combined approach. For the human dataset, the suggested method outperforms the CNN-BIGRU+ATT model by +1.0 % for cellular components, +1.1 % for molecular functions, and +0.5 % for biological processes. For the yeast dataset, the suggested method outperforms the CNN-BIGRU+ATT model by +2.4 % for the cellular component, +1.2 % for molecular functions, and +0.6 % for biological processes.

ICSDA: a multi-modal deep learning model to predict breast cancer recurrence and metastasis risk by integrating pathological, clinical and gene expression data.

Breast cancer patients often have recurrence and metastasis after surgery. Predicting the risk of recurrence and metastasis for a breast cancer patient is essential for the development of precision treatment. In this study, we proposed a novel multi-modal deep learning prediction model by integrating hematoxylin & eosin (H&E)-stained histopathological images, clinical information and gene expression data. Specifically, we segmented tumor regions in H&E into image blocks (256 × 256 pixels) and encoded each image block into a 1D feature vector using a deep neural network. Then, the attention module scored each area of the H&E-stained images and combined image features with clinical and gene expression data to predict the risk of recurrence and metastasis for each patient. To test the model, we downloaded all 196 breast cancer samples from the Cancer Genome Atlas with clinical, gene expression and H&E information simultaneously available. The samples were then divided into the training and testing sets with a ratio of 7: 3, in which the distributions of the samples were kept between the two datasets by hierarchical sampling. The multi-modal model achieved an area-under-the-curve value of 0.75 on the testing set better than those based solely on H&E image, sequencing data and clinical data, respectively. This study might have clinical significance in identifying high-risk breast cancer patients, who may benefit from postoperative adjuvant treatment.

Fine-grained image classification on bats using VGG16-CBAM: a practical example with 7 horseshoe bats taxa (CHIROPTERA: Rhinolophidae: Rhinolophus) from Southern China.

BACKGROUND: Rapid identification and classification of bats are critical for practical applications. However, species identification of bats is a typically detrimental and time-consuming manual task that depends on taxonomists and well-trained experts. Deep Convolutional Neural Networks (DCNNs) provide a practical approach for the extraction of the visual features and classification of objects, with potential application for bat classification. RESULTS: In this study, we investigated the capability of deep learning models to classify 7 horseshoe bat taxa (CHIROPTERA: Rhinolophus) from Southern China. We constructed an image dataset of 879 front, oblique, and lateral targeted facial images of live individuals collected during surveys between 2012 and 2021. All images were taken using a standard photograph protocol and setting aimed at enhancing the effectiveness of the DCNNs classification. The results demonstrated that our customized VGG16-CBAM model achieved up to 92.15% classification accuracy with better performance than other mainstream models. Furthermore, the Grad-CAM visualization reveals that the model pays more attention to the taxonomic key regions in the decision-making process, and these regions are often preferred by bat taxonomists for the classification of horseshoe bats, corroborating the validity of our methods. CONCLUSION: Our finding will inspire further research on image-based automatic classification of chiropteran species for early detection and potential application in taxonomy.

An Aerial Image Detection Algorithm Based on Improved YOLOv5.

To enhance aerial image detection in complex environments characterized by multiple small targets and mutual occlusion, we propose an aerial target detection algorithm based on an improved version of YOLOv5 in this paper. Firstly, we employ an improved Mosaic algorithm to address redundant boundaries arising from varying image scales and to augment the training sample size, thereby enhancing detection accuracy. Secondly, we integrate the constructed hybrid attention module into the backbone network to enhance the model's capability in extracting pertinent feature information. Subsequently, we incorporate feature fusion layer 7 and P2 fusion into the neck network, leading to a notable enhancement in the model's capability to detect small targets. Finally, we replace the original PAN + FPN network structure with the optimized BiFPN (Bidirectional Feature Pyramid Network) to enable the model to preserve deeper semantic information, thereby enhancing detection capabilities for dense objects. Experimental results indicate a substantial improvement in both the detection accuracy and speed of the enhanced algorithm compared to its original version. It is noteworthy that the enhanced algorithm exhibits a markedly improved detection performance for aerial images, particularly under real-time conditions.

EMA-YOLO: A Novel Target-Detection Algorithm for Immature Yellow Peach Based on YOLOv8.

Accurate determination of the number and location of immature small yellow peaches is crucial for bagging, thinning, and estimating yield in modern orchards. However, traditional methods have faced challenges in accurately distinguishing immature yellow peaches due to their resemblance to leaves and susceptibility to variations in shooting angles and distance. To address these issues, we proposed an improved target-detection model (EMA-YOLO) based on YOLOv8. Firstly, the sample space was enhanced algorithmically to improve the diversity of samples. Secondly, an EMA attention-mechanism module was introduced to encode global information; this module could further aggregate pixel-level features through dimensional interaction and strengthen small-target-detection capability by incorporating a 160 × 160 detection head. Finally, EIoU was utilized as a loss function to reduce the incidence of missed detections and false detections of the target small yellow peaches under the condition of high density of yellow peaches. Experimental results show that compared with the original YOLOv8n model, the EMA-YOLO model improves mAP by 4.2%, Furthermore, compared with SDD, Objectbox, YOLOv5n, and YOLOv7n, this model's mAP was improved by 30.1%, 14.2%,15.6%, and 7.2%, respectively. In addition, the EMA-YOLO model achieved good results under different conditions of illumination and shooting distance and significantly reduced the number of missed detections. Therefore, this method can provide technical support for smart management of yellow-peach orchards.

Enhanced Hybrid Vision Transformer with Multi-Scale Feature Integration and Patch Dropping for Facial Expression Recognition.

Convolutional neural networks (CNNs) have made significant progress in the field of facial expression recognition (FER). However, due to challenges such as occlusion, lighting variations, and changes in head pose, facial expression recognition in real-world environments remains highly challenging. At the same time, methods solely based on CNN heavily rely on local spatial features, lack global information, and struggle to balance the relationship between computational complexity and recognition accuracy. Consequently, the CNN-based models still fall short in their ability to address FER adequately. To address these issues, we propose a lightweight facial expression recognition method based on a hybrid vision transformer. This method captures multi-scale facial features through an improved attention module, achieving richer feature integration, enhancing the network's perception of key facial expression regions, and improving feature extraction capabilities. Additionally, to further enhance the model's performance, we have designed the patch dropping (PD) module. This module aims to emulate the attention allocation mechanism of the human visual system for local features, guiding the network to focus on the most discriminative features, reducing the influence of irrelevant features, and intuitively lowering computational costs. Extensive experiments demonstrate that our approach significantly outperforms other methods, achieving an accuracy of 86.51% on RAF-DB and nearly 70% on FER2013, with a model size of only 3.64 MB. These results demonstrate that our method provides a new perspective for the field of facial expression recognition.

CenterADNet: Infrared Video Target Detection Based on Central Point Regression.

Infrared video target detection is a fundamental technology within infrared warning and tracking systems. In long-distance infrared remote sensing images, targets often manifest as circular spots or even single points. Due to the weak and similar characteristics of the target to the background noise, the intelligent detection of these targets is extremely complex. Existing deep learning-based methods are affected by the downsampling of image features by convolutional neural networks, causing the features of small targets to almost disappear. So, we propose a new infrared video weak-target detection network based on central point regression. We focus on suppressing the image background by fusing the different features between consecutive frames with the original image features to eliminate the background's influence. We also employ high-resolution feature preservation and incorporate a spatial-temporal attention module into the network to capture as many target features as possible and improve detection accuracy. Our method achieves superior results on the infrared image weak aircraft target detection dataset proposed by the National University of Defense Technology, as well as on the simulated dataset generated based on real-world observation. This demonstrates the efficiency of our approach for detecting weak point targets in infrared continuous images.

Deep Recyclable Trash Sorting Using Integrated Parallel Attention.

Sorting recyclable trash is critical to reducing energy consumption and mitigating environmental pollution. Currently, trash sorting heavily relies on manpower. Computer vision technology enables automated trash sorting. However, existing trash image classification datasets contain a large number of images without backgrounds. Moreover, the models are vulnerable to background interference when categorizing images with complex backgrounds. In this work, we provide a recyclable trash dataset that supports model training and design a model specifically for trash sorting. Firstly, we introduce the TrashIVL dataset, an image dataset for recyclable trash sorting encompassing five classes (TrashIVL-5). All images are collected from public trash datasets, and the original images were captured by RGB imaging sensors, containing trash items with real-life backgrounds. To achieve refined recycling and improve sorting efficiency, the TrashIVL dataset can be further categorized into 12 classes (TrashIVL-12). Secondly, we propose the integrated parallel attention module (IPAM). Considering the susceptibility of sensor-based systems to background interference in real-world trash sorting scenarios, our IPAM is specifically designed to focus on the essential features of trash images from both channel and spatial perspectives. It can be inserted into convolutional neural networks (CNNs) as a plug-and-play module. We have constructed a recyclable trash sorting network building upon the IPAM, which produces an acuracy of 97.42% on TrashIVL-5 and 94.08% on TrashIVL-12. Our work is an effective attempt of computer vision in recyclable trash sorting. It makes a positive contribution to environmental protection and sustainable development.

EHNet: Efficient Hybrid Network with Dual Attention for Image Deblurring.

The motion of an object or camera platform makes the acquired image blurred. This degradation is a major reason to obtain a poor-quality image from an imaging sensor. Therefore, developing an efficient deep-learning-based image processing method to remove the blur artifact is desirable. Deep learning has recently demonstrated significant efficacy in image deblurring, primarily through convolutional neural networks (CNNs) and Transformers. However, the limited receptive fields of CNNs restrict their ability to capture long-range structural dependencies. In contrast, Transformers excel at modeling these dependencies, but they are computationally expensive for high-resolution inputs and lack the appropriate inductive bias. To overcome these challenges, we propose an Efficient Hybrid Network (EHNet) that employs CNN encoders for local feature extraction and Transformer decoders with a dual-attention module to capture spatial and channel-wise dependencies. This synergy facilitates the acquisition of rich contextual information for high-quality image deblurring. Additionally, we introduce the Simple Feature-Embedding Module (SFEM) to replace the pointwise and depthwise convolutions to generate simplified embedding features in the self-attention mechanism. This innovation substantially reduces computational complexity and memory usage while maintaining overall performance. Finally, through comprehensive experiments, our compact model yields promising quantitative and qualitative results for image deblurring on various benchmark datasets.

Optical Encryption Using Attention-Inserted Physics-Driven Single-Pixel Imaging.

Optical encryption based on single-pixel imaging (SPI) has made great advances with the introduction of deep learning. However, the use of deep neural networks usually requires a long training time, and the networks need to be retrained once the target scene changes. With this in mind, we propose an SPI encryption scheme based on an attention-inserted physics-driven neural network. Here, an attention module is used to encrypt the single-pixel measurement value sequences of two images, together with a sequence of cryptographic keys, into a one-dimensional ciphertext signal to complete image encryption. Then, the encrypted signal is fed into a physics-driven neural network for high-fidelity decoding (i.e., decryption). This scheme eliminates the need for pre-training the network and gives more freedom to spatial modulation. Both simulation and experimental results have demonstrated the feasibility and eavesdropping resistance of this scheme. Thus, it will lead SPI-based optical encryption closer to intelligent deep encryption.

Marine Oil Pollution Monitoring Based on a Morphological Attention U-Net Using SAR Images.

This study proposed an improved full-scale aggregated MobileUNet (FA-MobileUNet) model to achieve more complete detection results of oil spill areas using synthetic aperture radar (SAR) images. The convolutional block attention module (CBAM) in the FA-MobileUNet was modified based on morphological concepts. By introducing the morphological attention module (MAM), the improved FA-MobileUNet model can reduce the fragments and holes in the detection results, providing complete oil spill areas which were more suitable for describing the location and scope of oil pollution incidents. In addition, to overcome the inherent category imbalance of the dataset, label smoothing was applied in model training to reduce the model's overconfidence in majority class samples while improving the model's generalization ability. The detection performance of the improved FA-MobileUNet model reached an mIoU (mean intersection over union) of 84.55%, which was 17.15% higher than that of the original U-Net model. The effectiveness of the proposed model was then verified using the oil pollution incidents that significantly impacted Taiwan's marine environment. Experimental results showed that the extent of the detected oil spill was consistent with the oil pollution area recorded in the incident reports.

Improved MobileNet V3-Based Identification Method for Road Adhesion Coefficient.

To enable the timely adjustment of the control strategy of automobile active safety systems, enhance their capacity to adapt to complex working conditions, and improve driving safety, this paper introduces a new method for predicting road surface state information and recognizing road adhesion coefficients using an enhanced version of the MobileNet V3 model. On one hand, the Squeeze-and-Excitation (SE) is replaced by the Convolutional Block Attention Module (CBAM). It can enhance the extraction of features effectively by considering both spatial and channel dimensions. On the other hand, the cross-entropy loss function is replaced by the Bias Loss function. It can reduce the random prediction problem occurring in the optimization process to improve identification accuracy. Finally, the proposed method is evaluated in an experiment with a four-wheel-drive ROS robot platform. Results indicate that a classification precision of 95.53% is achieved, which is higher than existing road adhesion coefficient identification methods.

Optimal Channel Selection of Multiclass Motor Imagery Classification Based on Fusion Convolutional Neural Network with Attention Blocks.

The widely adopted paradigm in brain-computer interfaces (BCIs) involves motor imagery (MI), enabling improved communication between humans and machines. EEG signals derived from MI present several challenges due to their inherent characteristics, which lead to a complex process of classifying and finding the potential tasks of a specific participant. Another issue is that BCI systems can result in noisy data and redundant channels, which in turn can lead to increased equipment and computational costs. To address these problems, the optimal channel selection of a multiclass MI classification based on a Fusion convolutional neural network with Attention blocks (FCNNA) is proposed. In this study, we developed a CNN model consisting of layers of convolutional blocks with multiple spatial and temporal filters. These filters are designed specifically to capture the distribution and relationships of signal features across different electrode locations, as well as to analyze the evolution of these features over time. Following these layers, a Convolutional Block Attention Module (CBAM) is used to, further, enhance EEG signal feature extraction. In the process of channel selection, the genetic algorithm is used to select the optimal set of channels using a new technique to deliver fixed as well as variable channels for all participants. The proposed methodology is validated showing 6.41% improvement in multiclass classification compared to most baseline models. Notably, we achieved the highest results of 93.09% for binary classes involving left-hand and right-hand movements. In addition, the cross-subject strategy for multiclass classification yielded an impressive accuracy of 68.87%. Following channel selection, multiclass classification accuracy was enhanced, reaching 84.53%. Overall, our experiments illustrated the efficiency of the proposed EEG MI model in both channel selection and classification, showing superior results with either a full channel set or a reduced number of channels.

Gearbox Fault Diagnosis Based on MSCNN-LSTM-CBAM-SE.

Ensuring the safety of mechanical equipment, gearbox fault diagnosis is crucial for the stable operation of the whole system. However, existing diagnostic methods still have limitations, such as the analysis of single-scale features and insufficient recognition of global temporal dependencies. To address these issues, this article proposes a new method for gearbox fault diagnosis based on MSCNN-LSTM-CBAM-SE. The output of the CBAM-SE module is deeply integrated with the multi-scale features from MSCNN and the temporal features from LSTM, constructing a comprehensive feature representation that provides richer and more precise information for fault diagnosis. The effectiveness of this method has been validated with two sets of gearbox datasets and through ablation studies on this model. Experimental results show that the proposed model achieves excellent performance in terms of accuracy and F1 score, among other metrics. Finally, a comparison with other relevant fault diagnosis methods further verifies the advantages of the proposed model. This research offers a new solution for accurate fault diagnosis of gearboxes.

Detection of peach soluble solids based on near-infrared spectroscopy with High Order Spatial Interaction network.

BACKGROUND: Due to the scalability of deep learning technology, researchers have applied it to the non-destructive testing of peach internal quality. In addition, the soluble solids content (SSC) is an important internal quality indicator that determines the quality of peaches. Peaches with high SSC have a sweeter taste and better texture, making them popular in the market. Therefore, SSC is an important indicator for measuring peach internal quality and making harvesting decisions. RESULTS: This article presents the High Order Spatial Interaction Network (HOSINet), which combines the Position Attention Module (PAM) and Channel Attention Module (CAM). Additionally, a feature wavelength selection algorithm similar to the Group-based Clustering Subspace Representation (GCSR-C) is used to establish the Position and Channel Attention Module-High Order Spatial Interaction (PC-HOSI) model for peach SSC prediction. The accuracy of this model is compared with traditional machine learning and traditional deep learning models. Finally, the permutation algorithm is combined with deep learning models to visually evaluate the importance of feature wavelengths. Increasing the order of the PC-HOSI model enhances its ability to learn spatial correlations in the dataset, thus improving its predictive performance. CONCLUSION: The optimal model, PC-HOSI model, performed well with an order of 3 (PC-HOSI-3), with a root mean square error of 0.421 °Brix and a coefficient of determination of 0.864. Compared with traditional machine learning and deep learning algorithms, the coefficient of determination for the prediction set was improved by 0.07 and 0.39, respectively. The permutation algorithm also provided interpretability analysis for the predictions of the deep learning model, offering insights into the importance of spectral bands. These results contribute to the accurate prediction of SSC in peaches and support research on interpretability of neural network models for prediction. © 2024 Society of Chemical Industry.

Motor Fault Diagnosis Based on Convolutional Block Attention Module-Xception Lightweight Neural Network.

Electric motors play a crucial role in self-driving vehicles. Therefore, fault diagnosis in motors is important for ensuring the safety and reliability of vehicles. In order to improve fault detection performance, this paper proposes a motor fault diagnosis method based on vibration signals. Firstly, the vibration signals of each operating state of the motor at different frequencies are measured with vibration sensors. Secondly, the characteristic of Gram image coding is used to realize the coding of time domain information, and the one-dimensional vibration signals are transformed into grayscale diagrams to highlight their features. Finally, the lightweight neural network Xception is chosen as the main tool, and the attention mechanism Convolutional Block Attention Module (CBAM) is introduced into the model to enforce the importance of the characteristic information of the motor faults and realize their accurate identification. Xception is a type of convolutional neural network; its lightweight design maintains excellent performance while significantly reducing the model's order of magnitude. Without affecting the computational complexity and accuracy of the network, the CBAM attention mechanism is added, and Gram's corner field is combined with the improved lightweight neural network. The experimental results show that this model achieves a better recognition effect and faster iteration speed compared with the traditional Convolutional Neural Network (CNN), ResNet, and Xception networks.

[Multi-scale medical image segmentation based on pixel encoding and spatial attention mechanism].

In response to the issues of single-scale information loss and large model parameter size during the sampling process in U-Net and its variants for medical image segmentation, this paper proposes a multi-scale medical image segmentation method based on pixel encoding and spatial attention. Firstly, by redesigning the input strategy of the Transformer structure, a pixel encoding module is introduced to enable the model to extract global semantic information from multi-scale image features, obtaining richer feature information. Additionally, deformable convolutions are incorporated into the Transformer module to accelerate convergence speed and improve module performance. Secondly, a spatial attention module with residual connections is introduced to allow the model to focus on the foreground information of the fused feature maps. Finally, through ablation experiments, the network is lightweighted to enhance segmentation accuracy and accelerate model convergence. The proposed algorithm achieves satisfactory results on the Synapse dataset, an official public dataset for multi-organ segmentation provided by the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), with Dice similarity coefficient (DSC) and 95% Hausdorff distance (HD95) scores of 77.65 and 18.34, respectively. The experimental results demonstrate that the proposed algorithm can enhance multi-organ segmentation performance, potentially filling the gap in multi-scale medical image segmentation algorithms, and providing assistance for professional physicians in diagnosis.

[Feature detection of B-ultrasound images of intussusception in children based on improved YOLOv8n].

To assist grassroots sonographers in accurately and rapidly detecting intussusception lesions from children's abdominal ultrasound images, this paper proposes an improved YOLOv8n children's intussusception detection algorithm, called EMC-YOLOv8n. Firstly, the EfficientViT network with a cascaded group attention module was used as the backbone network to enhance the speed of target detection. Secondly, the improved C2fMBC module was used to replace the C2f module in the neck network to reduce network complexity, and the coordinate attention (CA) module was introduced after each C2fMBC module to enhance attention to positional information. Finally, experiments were conducted on the self-built dataset of intussusception in children. The results showed that the recall rate, average detection accuracy (mAP@0.5) and precision of the EMC-YOLOv8n algorithm improved by 3.9%, 2.1% and 0.9%, respectively, compared to the baseline algorithm. Despite slightly increased network parameters and computational load, significant improvements in detection accuracy enable efficient completion of detection tasks, demonstrating substantial economic and social value.

Attention-based Deep Learning for the Preoperative Differentiation of Axillary Lymph Node Metastasis in Breast Cancer on DCE-MRI.

BACKGROUND: Previous studies have explored the potential on radiomics features of primary breast cancer tumor to identify axillary lymph node (ALN) metastasis. However, the value of deep learning (DL) to identify ALN metastasis remains unclear. PURPOSE: To investigate the potential of the proposed attention-based DL model for the preoperative differentiation of ALN metastasis in breast cancer on dynamic contrast-enhanced MRI (DCE-MRI). STUDY TYPE: Retrospective. POPULATION: A total of 941 breast cancer patients who underwent DCE-MRI before surgery were included in the training (742 patients), internal test (83 patients), and external test (116 patients) cohorts. FIELD STRENGTH/SEQUENCE: A 3.0 T MR scanner, DCE-MRI sequence. ASSESSMENT: A DL model containing a 3D deep residual network (ResNet) architecture and a convolutional block attention module, named RCNet, was proposed for ALN metastasis identification. Three RCNet models were established based on the tumor, ALN, and combined tumor-ALN regions on the images. The performance of these models was compared with ResNet models, radiomics models, the Memorial Sloan-Kettering Cancer Center (MSKCC) model, and three radiologists (W.L., H.S., and F. L.). STATISTICAL TESTS: Dice similarity coefficient for breast tumor and ALN segmentation. Accuracy, sensitivity, specificity, intercorrelation and intracorrelation coefficients, area under the curve (AUC), and Delong test for ALN classification. RESULTS: The optimal RCNet model, that is, RCNet-tumor+ALN , achieved an AUC of 0.907, an accuracy of 0.831, a sensitivity of 0.824, and a specificity of 0.837 in the internal test cohort, as well as an AUC of 0.852, an accuracy of 0.828, a sensitivity of 0.792, and a specificity of 0.853 in the external test cohort. Additionally, with the assistance of RCNet-tumor+ALN , the radiologists' performance was improved (external test cohort, P < 0.05). DATA CONCLUSION: DCE-MRI-based RCNet model could provide a noninvasive auxiliary tool to identify ALN metastasis preoperatively in breast cancer, which may assist radiologists in conducting more accurate evaluation of ALN status. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 2.

Gastric polyp detection module based on improved attentional feature fusion.

Gastric cancer is a deadly disease and gastric polyps are at high risk of becoming cancerous. Therefore, the timely detection of gastric polyp is of great importance which can reduce the incidence of gastric cancer effectively. At present, the object detection method based on deep learning is widely used in medical images. However, as the contrast between the background and the polyps is not strong in gastroscopic image, it is difficult to distinguish various sizes of polyps from the background. In this paper, to improve the detection performance metrics of endoscopic gastric polyps, we propose an improved attentional feature fusion module. First, in order to enhance the contrast between the background and the polyps, we propose an attention module that enables the network to make full use of the target location information, it can suppress the interference of the background information and highlight the effective features. Therefore, on the basis of accurate positioning, it can focus on detecting whether the current location is the gastric polyp or background. Then, it is combined with our feature fusion module to form a new attentional feature fusion model that can mitigate the effects caused by semantic differences in the processing of feature fusion, using multi-scale fusion information to obtain more accurate attention weights and improve the detection performance of polyps of different sizes. In this work, we conduct experiments on our own dataset of gastric polyps. Experimental results show that the proposed attentional feature fusion module is better than the common feature fusion module and can improve the situation where polyps are missed or misdetected.