Pooling Operations Research Articles

Improved channel attention methods via hierarchical pooling and reducing information loss

Channel attention has been demonstrated to improve performance of convolutional neural networks. Most existing channel attention methods lower channel dimension for reducing computational complexity. However, the dimension reduction causes information loss, thus resulting in performance loss. To alleviate the paradox of complexity and performance trade-off, we propose two novel channel attention methods named Grouping-Shuffle-Aggregation Channel Attention (GSACA) method and Mixed Encoding Channel Attention (MECA) method, respectively. Our GSACA method partitions channel variables into several groups and performs the independent matrix multiplication without the dimension reduction to each group. Our GSACA method enables interaction between all groups using a ”channel shuffle” operator. After these, our GSACA method performs the independent matrix multiplication each group again and aggregates all channel correlations. Our MECA method encodes channel information through dual path architectures to benefit from both path topology where one uses the multilayer perception with dimension reduction to encode channel information and the other uses channel information encoding method without dimension reduction. Furthermore, a novel pooling operator named hierarchical pooling is presented and applied to our GSACA and MECA methods. The experimental results showed that our GSACA method almost consistently outperformed most existing channel attention methods and that our MECA method consistently outperformed the existing channel attention methods.

Research on image recognition of three Fritillaria cirrhosa species based on deep learning

Based on the deep learning method, a network model that can quickly and accurately identify the species of Fritillaria cirrhosa species was constructed. The learning method based on deep residual convolutional neural network was used to input the unprocessed original image directly as input, and the features of the image were extracted through convolution and pooling operations. On this basis, the ResNet34 model was improved, and the additional fully connected layer was added in front of the Softmax classifier to improve the learning ability of the network model. Total of 3915 images of three kinds of Fritillaria cirrhosa were used as data sources for the experiments, among which 160 images of each type were randomly selected to form the validation set. The final training set recognition accuracy rate was 95.8%, the validation set accuracy rate reached 92.3%, and the test set accuracy rate was 88.7%. The image recognition method of Fritillaria cirrhosa based on deep learning proposed in this paper is effective and feasible, which can quickly and accurately identify the species of Fritillaria cirrhosa species, and provides a new idea for the intelligent recognition of Chinese medicinal materials.

Movie recommendation system using taymon optimized deep learning network

The increased usage of the internet and social networks generates a large volume of information. Exploring through the large collection is time-consuming and hard to find the required one, so there is a serious need for a recommendation system. Based on this context several movie recommendation (MR) systems have been recently established. In addition, they have poor data analytics capability and cannot handle changing user preferences. As a result, there are many movies listed on the recommendation page, which provides for a poor user experience is the major issue. Therefore, in this work, a novel Taymon Optimized Deep Learning network (TODL net) for recommending top best movies based on their past choices, behaviour and movie contents. The deep neural network is a combination of Dilated CNN with Bi-directional LSTM. The DiCNN-BiLSTM model eliminates the functionality pooling operations and uses a dilated convolution layer to address the issue of information loss. The DiCNN is employed to learn the movie contents by mining user behavioral pattern attributes. The BiLSTM is applied to recommend the best movies on basis of the extracted features of the movie rating sequences of users in other social mediums. Moreover, for providing better results the DiCNN-BiLSTM is optimized with Taymon optimization algorithm to recommend best movies for the users. The proposed TODL net obtains the overall accuracy of 97.24% for best movies recommendation by using TMDB and MovieLens datasets.

MFAGNet: multi-scale frequency attention gating network for land cover classification

ABSTRACT As a classic problem of high-resolution remote sensing image tasks, land cover classification has many challenges, for instance, variable scales and low-edge discrimination. Traditional spatial semantic segmentation methods based on single-scale feature extraction are challenging to model multi-scale features. They cannot effectively fit high-resolution remote sensing images with variable scales and are prone to misclassification issues. On the other hand, spatial multi-scale semantic segmentation methods lack edge constraints, which can easily lead to edge segmentation ambiguity issues. We propose a segmentation network based on a multi-scale frequency-domain attention gating mechanism (MFAGNet) to solve the above problems. Specifically, we obtain multi-scale features through multi-scale input and design the discrete cosine transform channel attention module (DCTCAM). In DCTCAM, we extract global low-frequency features and effectively alleviate misclassification caused by scale changes. The edge features of the image are obtained through the high-frequency feature extraction module (HFEM) by using edge detection operators to enhance the model’s ability to recognize edges. Additionally, no further pooling operations are employed except for a single max pooling layer in the backbone network. In this paper, we design a new gating mechanism to dynamically control the extraction ratio of various frequency features so that it can adapt to input images of different scales. This paper conducts comparative experiments on three high-resolution land cover datasets: GID, Vaihingen, and Potsdam. The new algorithm proposed compares with classic semantic segmentation models such as DANet, PSPNet, SegNet, UNet, DeepLabv3+, FcaNet, BiSeNet, AFN, and FastFCN. The experimental results show that MFAGNet is superior to the comparison models. With no significant increase in computing overhead, MFAGNet obtains MIoU of 69.05%, 73.40%, and 79.42% on the standard data sets GID, Vaihingen, and Potsdam, respectively.

Deep learning based hybrid prediction model for predicting the spread of COVID-19 in the world's most populous countries.

COVID-19 has a disease and health phenomenon and has sociological and economic adverse effects. Accurate prediction of the spread of the epidemic will help in the planning of health management and the development of economic and sociological action plans. In the literature, there are many studies to analyse and predict the spread of COVID-19 in cities and countries. However, there is no study to predict and analyse the cross-country spread in the world's most populous countries. In this study, it was aimed to predict the spread of the COVID-19 epidemic. The motivation of this study is to reduce the workload of health workers, take preventive measures and optimize health processes by predicting the spread of the COVID-19 epidemic. A hybrid deep learning model was developed to predict and analyse COVID-19 cross-country spread and a case study was carried out for the world's most populous countries. The developed model was tested extensively using RMSE, MAE and R2. The experimental results showed that the developed model was more successful in predicting and analysis of COVID-19 cross-country spread in the world's most populous countries than LR, RF, SVM, MLP, CNN, GRU, LSTM and base CNN-GRU. In the developed model, CNN performs convolution and pooling operations to extract spatial features from the input data. GRU provides learning of long-term and non-linear relationships inferred by CNN. The developed hybrid model was more successful than the other models compared, as it enabled the effective features of the CNN and GRU models to be used together. The prediction and analysis of the cross-country spread of COVID-19 in the world's most populated countries can be presented as a novelty of this study.

Classification of urban interchange patterns using a model combining shape context descriptor and graph convolutional neural network

ABSTRACT Pattern recognition is critical to map data handling and their applications. This study presents a model that combines the Shape Context (SC) descriptor and Graph Convolutional Neural Network (GCNN) to classify the patterns of interchanges, which are indispensable parts of urban road networks. In the SC-GCNN model, an interchange is modeled as a graph, wherein nodes and edges represent the interchange segments and their connections, respectively. Then, a novel SC descriptor is implemented to describe the contextual information of each interchange segment and serve as descriptive features of graph nodes. Finally, a GCNN is designed by combining graph convolution and pooling operations to process the constructed graphs and classify the interchange patterns. The SC-GCNN model was validated using interchange samples obtained from the road networks of 15 cities downloaded from OpenStreetMap. The classification accuracy was 87.06%, which was higher than that of the image-based AlexNet, GoogLeNet, and Random Forest models.

DenseNet-random forest model based galaxy classification

Finding an efficient and accurate adaptive method that can automatically classify galaxies has become an industry consensus. However, most of the current studies on galaxy classification use a single model for direct output, without considering the combination with other models to output more satisfactory prediction results. Through convolutional neural network and classifier, this study studied the possibility of applying the deep learning model to the Galaxy 10 DECals dataset classification, and proposed DenseNet-Random Forest model through comparative analysis. By adjusting and training DenseNet-121 with appropriate hyperparameters, the input tensor is transferred to the basic model through the creation of a shape input layer, where GlobalAveragePooling2D is added to perform an average pooling operation on each feature map, reducing the spatial dimension of each feature map to 1. During the process, a complete connection layer with 64 neurons was added using the ReLU activation function, and a Dropout layer was added to randomly discard 20% of the neurons during training to prevent overfitting. In addition, ReLU Activation function with 32 full connection layers of neurons and softmax Activation function with 10 output layers of neurons are added. By acquiring the feature vector of the training model and the real label of the verification set, assign x and y values respectively, and import them into the Random forest classifier model. The experimental results demonstrated the model ultimately achieved a prediction accuracy of 68% when processing the Galaxy 10 DECals dataset, and achieved nearly 30% improvement in Precision, Recall, and F1 scores.

An Efficient and Accurate Convolution-Based Similarity Measure for Uncertain Trajectories

With the rapid development of localization techniques and the prevalence of mobile devices, massive amounts of trajectory data have been generated, playing essential roles in areas of user analytics, smart transportation, and public safety. Measuring trajectory similarity is one of the fundamental tasks in trajectory analytics. Although considerable research has been conducted on trajectory similarity, the majority of existing approaches measure the similarity between two trajectories by calculating the distance between aligned locations, leading to challenges related to uncertain trajectories (e.g., low and heterogeneous data sampling rates, as well as location noise). To address these challenges, we propose Contra, a convolution-based similarity measure designed specifically for uncertain trajectories. The main focus of Contra is to identify the similarity of trajectory shapes while disregarding the time/order relevance of each record within the trajectory. To this end, it leverages a series of convolution and pooling operations to extract high-level geo-information from trajectories, and subsequently compares their similarities based on these extracted features. Moreover, we introduce efficient trajectory index strategies to enhance the computational efficiency of our proposed measure. We conduct comprehensive experiments on two trajectory datasets to evaluate the performance of our proposed approach. The experiments on both datasets show the effectiveness and efficiency of our approach. Specifically, the mean rank of Contra is 3 times better than the state-of-the-art approaches, and the precision of Contra surpasses baseline approaches by 20–40%.

TransPCGC: Point Cloud Geometry Compression Based on Transformers

Due to the often substantial size of the real-world point cloud data, efficient transmission and storage have become critical concerns. Point cloud compression plays a decisive role in addressing these challenges. Recognizing the importance of capturing global information within point cloud data for effective compression, many existing point cloud compression methods overlook this crucial aspect. To tackle this oversight, we propose an innovative end-to-end point cloud compression method designed to extract both global and local information. Our method includes a novel Transformer module to extract rich features from the point cloud. Utilization of a pooling operation that requires no learnable parameters as a token mixer for computing long-distance dependencies ensures global feature extraction while significantly reducing both computations and parameters. Furthermore, we employ convolutional layers for feature extraction. These layers not only preserve the spatial structure of the point cloud, but also offer the advantage of parameter independence from the input point cloud size, resulting in a substantial reduction in parameters. Our experimental results demonstrate the effectiveness of the proposed TransPCGC network. It achieves average Bjontegaard Delta Rate (BD-Rate) gains of 85.79% and 80.24% compared to Geometry-based Point Cloud Compression (G-PCC). Additionally, in comparison to the Learned-PCGC network, our approach attains an average BD-Rate gain of 18.26% and 13.83%. Moreover, it is accompanied by a 16% reduction in encoding and decoding time, along with a 50% reduction in model size.

Fine‐grained crack segmentation for high‐resolution images via a multiscale cascaded network

AbstractHigh‐resolution (HR) crack images offer more detailed information for assessing structural conditions compared to low‐resolution (LR) images. This wealth of detail proves indispensable in bolstering the safety of unmanned aerial vehicle (UAV)‐based inspection procedures and elevating the precision of small crack segmentation. Nonetheless, achieving a balance between segmentation accuracy and GPU memory consumption poses a substantial challenge for deep learning models when processing HR crack images. To overcome this challenge, a novel “HR crack segmentation framework” (HRCSF) is proposed, specifically designed to meticulously segment crack images with resolutions exceeding 4K. First, a multiscale crack feature extraction network (MsCFEN) was proposed with the embedment of the strip pooling operation to enhance the representation of the transverse and longitudinal crack pixels from the complex backgrounds. Subsequently, two cascaded operations were tailored to MsCFEN, enabling a comprehensive refinement process that incorporates both global and local aspects. Furthermore, to fully leverage the potential of each proposed component in the refinement process, the complete architecture was trained using a loss function with embedded boundary optimization. Conclusively, a UAV‐based case study was conducted on a real bridge in Changsha, demonstrating HRCSF's practicability in segmenting HR crack images. The implementation of HRCSF allows the UAV to perform crack inspection effectively from a distance of 3 m away from the girder, resulting in a significant 50% reduction in inspection time compared to LR segmentation methods while maintaining high detection accuracy.

RMAU-Net: Breast Tumor Segmentation Network Based on Residual Depthwise Separable Convolution and Multiscale Channel Attention Gates

Breast cancer is one of the most common female diseases, posing a great threat to women’s health, and breast ultrasound imaging is a common method for breast cancer diagnosis. In recent years, U-Net and its variants have dominated the medical image segmentation field with their excellent performance. However, the existing U-type segmentation networks have the following problems: (1) the design of the feature extractor is complicated, and the calculation difficulty is increased; (2) the skip connection operation simply combines the features of the encoder and the decoder, without considering both spatial and channel dimensions; (3) during the downsampling phase, the pooling operation results in the loss of feature information. To address the above deficiencies, this paper proposes a breast tumor segmentation network, RMAU-Net, that combines residual depthwise separable convolution and a multi-scale channel attention gate. Specifically, we designed the RDw block, which has a simple structure and a larger sensory field, to overcome the localization problem of convolutional operations. Meanwhile, the MCAG module is designed to correct the low-level features in both spatial and channel dimensions and assist the high-level features to recover the up-sampling and pinpoint non-regular breast tumor features. In addition, this paper used the Patch Merging operation instead of the pooling method to prevent the loss of breast ultrasound image information. Experiments were conducted on two breast ultrasound datasets, Dataset B and BUSI, and the results show that the method in this paper has superior segmentation performance and better generalization.

Capturing the diffusive behavior of the multiscale linear transport equations by Asymptotic-Preserving Convolutional DeepONets

In this paper, we introduce two types of novel Asymptotic-Preserving Convolutional Deep Operator Networks (APCONs) designed to solve the multiscale time-dependent linear transport equations. We observe that the vanilla physics-informed DeepONets with modified MLP may exhibit instability in maintaining the desired limiting macroscopic behavior. Therefore, this necessitates the utilization of an asymptotic-preserving loss function. Drawing inspiration from the heat kernel in the diffusion equation, we propose a new architecture called Convolutional Deep Operator Networks, which employs multiple local convolution operations instead of a global heat kernel, along with pooling and activation operations in each filter layer. Our APCON methods possess a parameter count that is independent of the grid size and are capable of capturing the diffusive behavior of the linear transport problem. Finally, we validate the effectiveness of our methods through several numerical examples.

Lightweight Single Shot Multi-Box Detector: A fabric defect detection algorithm incorporating parallel dilated convolution and dual channel attention

For the textile industry, fabric defect detection is an important part of production. In order to make the automatic fabric defect detection system used in production sites, this article proposes a lightweight algorithm Lightweight Single Shot Multi-Box Detector (LW-SSD) to address the issues of low detection accuracy, high computational complexity, and difficulty in deploying on hardware devices with limited computing power in fabric defect detection. Firstly, MobileNetv3 is introduced as the backbone network to reduce the number of model parameters. Secondly, in the feature fusion module, down-sampling stacking is used to fuse the feature maps processed by maximum pooling and regular 3 × 3 convolution, respectively, to enhance the generalization and small target feature extraction capability of the network. Then, the dilated convolution is incorporated into the Inceptionv3 to form a multi-branch parallel dilated convolution module, which can expand the receptive field of the feature layer and enhance the extraction of the target information. Finally, a dual-channel attention module is added, which adds the maximum pooling operation based on the efficient channel attention for deep convolutional neural networks (ECA) channel attention mechanism to highlight defect features and suppress background noise features. The experiments show that the accuracy of the system is improved while maintaining the faster detection speed. Among them, the LW-SSD algorithm has an accuracy improvement of 10.03% on the self-made dataset, a reduction of 58% in the number of model parameters compared to the Single Shot Multi-Box Detector (SSD) algorithm, and the detection speed reaches 48 frames per second.

An efficient astronomical seeing forecasting method by random convolutional Kernel transformation

Astronomical seeing is the primary indicator of the observation quality of large-scale optical telescopes. Forecasting astronomical seeing is an essential and challenging task in the operation of large optical telescopes, as it can help observers more efficiently arrange observation tasks as well as providing hints for mechanism optimization of optical telescopes. In this paper, we propose an efficient data-driven astronomical seeing forecasting method by leveraging the environment monitoring data from multiple sensors deployed with optical telescopes. The basic idea is to use a large number of random yet fixed convolution kernels to extract features from the original data, and then train a simple tree-based predictor based on these features. In particular, the raw monitoring data is transformed by differential operations, enabling a more explicit representation of spatial and temporal features; then an elaborate convolution mechanism, including the convolution kernel and the pooling operator, is designed to extract a variety set of features of the input time series, which serve as the foundation for constructing the time series forecasting model. In contrast to existing deep learning models, there is no need to calculate and back-propagate the gradients during the training procedure, making the proposed method much more efficient. We collect real-world monitoring data from Large Sky Area Multi-Object Fiber Spectroscopy Telescope (LAMOST), one of the world’s largest optical reflecting telescopes, and conduct extensive empirical evaluations. The experimental results demonstrate that the proposed method can achieve state-of-the-art prediction accuracy, i.e., 0.1125 in MAE, 0.0252 in MSE, and 0.0322 in MAPE, outperforming most deep-learning-based forecasting models. Moreover, it surpasses both machine learning and deep learning approaches in terms of training efficiency with a significant margin.

DMCA-GAN: Dual Multilevel Constrained Attention GAN for MRI-Based Hippocampus Segmentation.

Precise segmentation of the hippocampus is essential for various human brain activity and neurological disorder studies. To overcome the small size of the hippocampus and the low contrast of MR images, a dual multilevel constrained attention GAN for MRI-based hippocampus segmentation is proposed in this paper, which is used to provide a relatively effective balance between suppressing noise interference and enhancing feature learning. First, we design the dual-GAN backbone to effectively compensate for the spatial information damage caused by multiple pooling operations in the feature generation stage. Specifically, dual-GAN performs joint adversarial learning on the multiscale feature maps at the end of the generator, which yields an average Dice coefficient (DSC) gain of 5.95% over the baseline. Next, to suppress MRI high-frequency noise interference, a multilayer information constraint unit is introduced before feature decoding, which improves the sensitivity of the decoder to forecast features by 5.39% and effectively alleviates the network overfitting problem. Then, to refine the boundary segmentation effects, we construct a multiscale feature attention restraint mechanism, which forces the network to concentrate more on effective multiscale details, thus improving the robustness. Furthermore, the dual discriminators D1 and D2 also effectively prevent the negative migration phenomenon. The proposed DMCA-GAN obtained a DSC of 90.53% on the Medical Segmentation Decathlon (MSD) dataset with tenfold cross-validation, which is superior to the backbone by 3.78%.

Dynamic Clustering Strategies Boosting Deep Learning in Olive Leaf Disease Diagnosis

Artificial intelligence has many applications in various industries, including agriculture. It can help overcome challenges by providing efficient solutions, especially in the early stages of development. When working with tree leaves to identify the type of disease, diseases often show up through changes in leaf color. Therefore, it is crucial to improve the color brightness before using them in intelligent agricultural systems. Color improvement should achieve a balance where no new colors appear, as this could interfere with accurate identification and diagnosis of the disease. This is considered one of the challenges in this field. This work proposes an effective model for olive disease diagnosis, consisting of five modules: image enhancement, feature extraction, clustering, and deep neural network. In image enhancement, noise reduction, balanced colors, and CLAHE are applied to LAB color space channels to improve image quality and visual stimulus. In feature extraction, raw images of olive leaves are processed through triple convolutional layers, max pooling operations, and flattening in the CNN convolutional phase. The classification process starts by dividing the data into clusters based on density, followed by the use of a deep neural network. The proposed model was tested on over 3200 olive leaf images and compared with two deep learning algorithms (VGG16 and Alexnet). The results of accuracy and loss rate show that the proposed model achieves (98%, 0.193), while VGG16 and Alexnet reach (96%, 0.432) and (95%, 1.74), respectively. The proposed model demonstrates a robust and effective approach for olive disease diagnosis that combines image enhancement techniques and deep learning-based classification to achieve accurate and reliable results.

DMNet: Dynamic Memory Network for RGB-D Salient Object Detection

RGB-D salient object detection aims to identify the most salient object using the color and depth information of images. Currently, most of the existing salient object detection models use the classical U-Net architecture, which uses up-sampling and short link decoding to exploit saliency cues after generating multi-level features by successive convolution and pooling operations. However, the multi-level encoding and progressive up-sampling decoding used by these models may lose some of the high-level semantic information during feature extraction and information fusion and some of the detailed information during image recovery, which in turn affects the quality of generated saliency maps. To solve these problems, we propose a dynamic memory network (DMNet) consisting of an interactive enhancement encoder and a dynamic memory decoder, which achieves high-precision detection of salient objects with a more complete and superior encoding and decoding strategy. Specifically, in the interactive enhancement encoder, we propose a feature interactive fusion module (FIFM), which can enhance intermediate-scale RGB features by fusing RGB features of three adjacent scales. Moreover, we design a depth-guided dense fusion module (DDFM) to fuse the features from RGB and depth modalities. In the dynamic memory decoder, we design a full-dimensional dynamic convolutional expansion module (FDCEM) to enhance the information representation capability of features at different scales. Furthermore, we also design a gated decoding module (GDM) to decode features of different scales and eliminate non-salient information. Extensive experimental results on STERE, SIP, NLPR, NJU2K, SSD, and DES datasets demonstrate that our model outperforms most of the state-of-the-art RGB-D SOD methods. In addition, the experimental results on three RGB-T datasets, VT821, VT1000, and VT5000 also show that our model can be effectively used for RGB-T SOD.

SFPFusion: An Improved Vision Transformer Combining Super Feature Attention and Wavelet-Guided Pooling for Infrared and Visible Images Fusion.

The infrared and visible image fusion task aims to generate a single image that preserves complementary features and reduces redundant information from different modalities. Although convolutional neural networks (CNNs) can effectively extract local features and obtain better fusion performance, the size of the receptive field limits its feature extraction ability. Thus, the Transformer architecture has gradually become mainstream to extract global features. However, current Transformer-based fusion methods ignore the enhancement of details, which is important to image fusion tasks and other downstream vision tasks. To this end, a new super feature attention mechanism and the wavelet-guided pooling operation are applied to the fusion network to form a novel fusion network, termed SFPFusion. Specifically, super feature attention is able to establish long-range dependencies of images and to fully extract global features. The extracted global features are processed by wavelet-guided pooling to fully extract multi-scale base information and to enhance the detail features. With the powerful representation ability, only simple fusion strategies are utilized to achieve better fusion performance. The superiority of our method compared with other state-of-the-art methods is demonstrated in qualitative and quantitative experiments on multiple image fusion benchmarks.

Enhancing the accuracies by performing pooling decisions adjacent to the output layer

Learning classification tasks of ({2}^{n}times {2}^{n}) inputs typically consist of le n(2times 2) max-pooling (MP) operators along the entire feedforward deep architecture. Here we show, using the CIFAR-10 database, that pooling decisions adjacent to the last convolutional layer significantly enhance accuracies. In particular, average accuracies of the advanced-VGG with m layers (A-VGGm) architectures are 0.936, 0.940, 0.954, 0.955, and 0.955 for m = 6, 8, 14, 13, and 16, respectively. The results indicate A-VGG8’s accuracy is superior to VGG16’s, and that the accuracies of A-VGG13 and A-VGG16 are equal, and comparable to that of Wide-ResNet16. In addition, replacing the three fully connected (FC) layers with one FC layer, A-VGG6 and A-VGG14, or with several linear activation FC layers, yielded similar accuracies. These significantly enhanced accuracies stem from training the most influential input–output routes, in comparison to the inferior routes selected following multiple MP decisions along the deep architecture. In addition, accuracies are sensitive to the order of the non-commutative MP and average pooling operators adjacent to the output layer, varying the number and location of training routes. The results call for the reexamination of previously proposed deep architectures and their accuracies by utilizing the proposed pooling strategy adjacent to the output layer.

MAS-Net:Multi-modal Assistant Segmentation Network For Lumbar Intervertebral Disc.

Objective.Despite advancements in medical imaging technology, the diagnosis and positioning of lumbar disc diseases still heavily rely on the expertise and experience of medical professionals. This process is often time-consuming, labor-intensive, and susceptible to subjective factors. Achieving automatic positioning and segmentation of lumbar intervertebral disc (LID) is the first and critical step in intelligent diagnosis of lumbar disc diseases. However, due to the complexity of the vertebral body and the ambiguity of the soft tissue boundaries of the LID, accurate and intelligent segmentation of LIDs remains challenging. The study aims to accurately and intelligently segment and locate LIDs by fully utilizing multi-modal lumbar magnetic resonance Images (MRIs).Approach.A novel multi-modal assistant segmentation network (MAS-Net) is proposed in this paper. The architecture consists of four key components: the multi-branch fusion encoder (MBFE), the cross-modality correlation evaluation (CMCE), the channel fusion transformer (CFT), and the selective Kernel (SK) based decoder. The MBFE module captures and integrates various modal features, while the CMCE module facilitates the fusion process between the MBFE and decoder. The CFT module selectively guides the flow of information between the MBFE and decoder and effectively utilizes skip connections from multiple layers. The SK module computes the significance of each channel using global pooling operations and applies weights to the input feature maps to improve the models recognition of important features.Main results.The proposed MAS-Net achieved a dice coefficient of 93.08% on IVD3Seg and 93.22% on DualModalDisc dataset, outperforming the current state-of-the-art network, accurately segmenting the LIDs, and generating a 3D model that can precisely display the LIDs.Significance.MAS-Net automates the diagnostics process and addresses challenges faced by doctors. Simplifying and enhancing the clarity of visual representation, multi-modal MRI allows for better information complementation and LIDs segmentation. By successfully integrating data from various modalities, the accuracy of LID segmentation is improved.