Lightweight Convolutional Network Research Articles

Lightweight Convolutional Network with Integrated Attention Mechanism for Missing Bolt Detection in Railways

Railway infrastructure safety is a paramount concern, with bolt integrity being a critical component. In the realm of railway maintenance, the detection of missing bolts is a vital task that ensures the stability and safety of tracks. Traditionally, this task has been approached through manual inspections or conventional automated methods, which are often time-consuming, costly, and prone to human error. Addressing these challenges, this paper presents a state-of-the-art solution with the development of a lightweight convolutional neural network (CNN) featuring an integrated attention mechanism. This novel model is engineered to be computationally efficient while maintaining high accuracy, making it particularly suitable for real-time analysis in resource-constrained environments commonly found in railway inspections. The proposed CNN utilises a distinctive architecture that synergises the speed of lightweight networks with the precision of attention-based mechanisms. By integrating an attention mechanism, the network selectively concentrates on regions of interest within the image, effectively enhancing the model’s capability to identify missing bolts with remarkable accuracy. Comprehensive testing showcases a remarkable 96.43% accuracy and an impressive 96 F1-score, substantially outperforming existing deep learning frameworks in the context of missing bolt detection. Key contributions of this research include the model’s innovative attention-integrated approach, which significantly reduces the model complexity without compromising detection performance. Additionally, the model offers scalability and adaptability to various railway settings, proving its efficacy not just in controlled environments but also in diverse real-world scenarios. Extensive experiments, rigorous evaluations, and real-time deployment results collectively underscore the transformative potential of the presented CNN model in advancing the domain of railway safety maintenance.

LightNet: a novel lightweight convolutional network for brain tumor segmentation in healthcare.

Diagnosis, treatment planning, surveillance, and the monitoring of clinical trials for brain diseases all benefit greatly from neuroimaging-based tumor segmentation. Recently, Convolutional Neural Networks (CNNs) have demonstrated promising results in enhancing the efficiency of image-based brain tumor segmentation. Most current work on CNNs, however, is devoted to creating increasingly complicated convolution modules to improve performance, which in turn raises the computing cost of the model. This work proposes a simple and effective feed-forward CNN, LightNet (Light Network). Based on multi-path and multi-level, it replaces traditional convolutional methods with light operations, which reduces network parameters and redundant feature maps. In the up-sampling stage, a light channel attention module is added to achieve richer multi-scale and spatial semantic feature information extraction of brain tumor. The performance of the network is evaluated in the Multimodal Brain Tumor Segmentation Challenge (BraTS 2015) dataset, and results are presented here alongside other high-performing CNNs. Results show comparable accuracy with other methods but with increased efficiency, segmentation performance, and reduced redundancy and computational complexity. The result is a high-performing network with a balance between efficiency and accuracy, allowing, for example, better energy performance on mobile devices.

Fast Context-Awareness Encoder for LiDAR Point Semantic Segmentation

A LiDAR sensor is a valuable tool for environmental perception as it can generate 3D point cloud data with reflectivity and position information by reflecting laser beams. However, it cannot provide the meaning of each point cloud cluster, so many studies focus on identifying semantic information about point clouds. This paper explores point cloud segmentation and presents a lightweight convolutional network called Fast Context-Awareness Encoder (FCAE), which can obtain semantic information about the point cloud cluster at different levels. The surrounding features of points are extracted as local features through the local context awareness network, then combined with global features, which are highly abstracted from the local features, to obtain more accurate semantic segmentation of the discrete points in space. The proposed algorithm has been compared and verified against other semantic KITTI data algorithms and has achieved state-of-the-art performance. Due to its ability to note fine-grained features on the z-axis in space, the algorithm shows higher prediction accuracy for certain types of objects. Moreover, the training and validation time is short, and the algorithm can meet high real-time requirements for 3D perception tasks.

IPAD: Iterative pruning with activation deviation for sclera biometrics

The sclera has recently been gaining attention as a biometric modality due to its various desirable characteristics. A key step in any type of ocular biometric recognition, including sclera recognition, is the segmentation of the relevant part(s) of the eye. However, the high computational complexity of the (deep) segmentation models used in this task can limit their applicability on resource-constrained devices such as smartphones or head-mounted displays. As these devices are a common desired target for such biometric systems, lightweight solutions for ocular segmentation are critically needed. To address this issue, this paper introduces IPAD (Iterative Pruning with Activation Deviation), a novel method for developing lightweight convolutional networks, that is based on model pruning. IPAD uses a novel filter-activation-based criterion (ADC) to determine low-importance filters and employs an iterative model pruning procedure to derive the final lightweight model. To evaluate the proposed pruning procedure, we conduct extensive experiments with two diverse segmentation models, over four publicly available datasets (SBVPI, SLD, SMD and MOBIUS), in four distinct problem configurations and in comparison to state-of-the-art methods from the literature. The results of the experiments show that the proposed filter-importance criterion outperforms the standard L1 and L2 approaches from the literature. Furthermore, the results also suggest that: (i) the pruned models are able to retain (or even improve on) the performance of the unpruned originals, as long as they are not over-pruned, with RITnet and U-Net at 50% of their original FLOPs reaching up to 4% and 7% higher IoU values than their unpruned versions, respectively, (ii) smaller models require more careful pruning, as the pruning process can hurt the model’s generalization capabilities, and (iii) the novel criterion most convincingly outperforms the classic approaches when sufficient training data is available, implying that the abundance of data leads to more robust activation-based importance computation.

LCH: fast RGB-D salient object detection on CPU via lightweight convolutional network with hybrid knowledge distillation

LCH: fast RGB-D salient object detection on CPU via lightweight convolutional network with hybrid knowledge distillation

Video based behavior detection algorithm

Action recognition based on video surveillance becomes possible because of the rapid development of action recognition, temporal action recognition and spatial-temporal action recognition technology. A video-based behavior detection algorithm designed to find information of interest from videos. In the process of video detection, feature extraction is often carried out from space and time dimensions. However, the calculation amount of videos sent into the deep convolutional network is much higher than that of pictures. Therefore, the design of lightweight convolutional network is conducive to reducing training costs, realizing efficient calculation, and enabling researchers to quickly apply it to practical scenarios.

Efficient and Lightweight Convolutional Networks for IoT Malware Detection: A Federated Learning Approach

Over the past few years, billions of unsecured Internet of Things (IoT) devices have been produced and released, and that number will only grow as wireless technology advances. As a result of their susceptibility to malware, effective methods have become necessary for identifying IoT malware. However, the low generalizability and the nonindependently and identically distributed data (non-IID) still pose a major challenge to achieving this goal. In this work, a new federated malware detection paradigm, termed FED-MAL, is introduced to collaboratively train multiple distributed edge devices to detect malware. In FED-MAL, the malware binaries are transformed into an image format to lessen the impact on non-IID, and then a compact convolutional model, named AM-NET, is proposed to learn the malware patterns as an image recognition task. The compact nature of AM-NET makes it an appropriate choice for deployment on resource-constrained IoT devices. Following, a refined edge-based adversarial training is given in FED-MAL to empower generalizability and resistibility by generating adversarial samples from various participating clients. Experimental evaluation on publicly available malware data sets shows that the FED-MAL is efficacious, reliable, expandable, generalizable, and communication efficient.

LUVS-Net: A Lightweight U-Net Vessel Segmentor for Retinal Vasculature Detection in Fundus Images

This paper presents LUVS-Net, which is a lightweight convolutional network for retinal vessel segmentation in fundus images that is designed for resource-constrained devices that are typically unable to meet the computational requirements of large neural networks. The computational challenges arise due to low-quality retinal images, wide variance in image acquisition conditions and disparities in intensity. Consequently, the training of existing segmentation methods requires a multitude of trainable parameters for the training of networks, resulting in computational complexity. The proposed Lightweight U-Net for Vessel Segmentation Network (LUVS-Net) can achieve high segmentation performance with only a few trainable parameters. This network uses an encoder–decoder framework in which edge data are transposed from the first layers of the encoder to the last layer of the decoder, massively improving the convergence latency. Additionally, LUVS-Net’s design allows for a dual-stream information flow both inside as well as outside of the encoder–decoder pair. The network width is enhanced using group convolutions, which allow the network to learn a larger number of low- and intermediate-level features. Spatial information loss is minimized using skip connections, and class imbalances are mitigated using dice loss for pixel-wise classification. The performance of the proposed network is evaluated on the publicly available retinal blood vessel datasets DRIVE, CHASE_DB1 and STARE. LUVS-Net proves to be quite competitive, outperforming alternative state-of-the-art segmentation methods and achieving comparable accuracy using trainable parameters that are reduced by two to three orders of magnitude compared with those of comparative state-of-the-art methods.

Automatic extraction of surface water based on lightweight convolutional neural network

Automatic extraction of surface water is of great significance to the study of the global water cycle and the dynamic management of water resources. At present, the water extraction accuracy of high-resolution multi-spectral remote sensing images has been greatly improved. But it is still affected by the shadow of the mountains and the shadow of the tall buildings inside the city. The spectral information of shadow is basically consistent with the spectral information of water, so the accuracy of any traditional index extraction of water will be questioned by users. Or the user must adjust threshold parameters many times to obtain good extraction results, which is contradictory to fast and large-area remote sensing monitoring. To solve the above problems, this paper firstly introduces the thermal infrared band at the data end for pre-treatment. Then, a lightweight neural network (EDCM) combining the most advanced lightweight image classification model and semantic segmentation model is proposed, which is specially used for fast, large area and automatic water extraction. Multi-scale training of samples using lightweight convolutional networks aims to extract multi-scale context information. The newly constructed model was tested in three highly heterogeneous scenarios, and the results showed that the trained EDCM model achieved the highest accuracy in all selected test areas, reaching more than 95.28%. The EDCM model can be used for the high-precision extraction of surface water in complex areas.

Time-frequency fusion features-based incremental network for smartphone measured structural seismic response classification

Smartphone as a portable device can be used for public participatory structural seismic response acquisition. This data acquisition approach not only reduces the acquisition cost but also facilitates the possibility to obtain city-scale building seismic responses. In this paper, we propose a novel structural seismic response classification method using time–frequency fusion features-based incremental network, namely SeismicNet, that can automatically classify the signals acquired through smartphones into structural response under normal operation conditions (structural normal response) or structural response under strong ground motion (structural seismic response). In this method, the time-domain signals of structural response are converted into time–frequency maps, which will be fed into a lightweight convolutional network for the extraction of the initial fusion features. Subsequently, these features are mapped by random weights to dynamic nodes. This not only improves the capability of feature expression, but also allows incremental updating of the model. Finally, a simulated structural response dataset and some real structural seismic responses are used to verify the accuracy and practicability of the proposed method. Furthermore, the results showed that the proposed method is more than 84 times in training speed and 2 times in inference speed than some well-known deep learning networks (VGG16 and ResNet50).

FLC-NET: Federated Lightweight Network for Early Discovery of Malware in Resource-constrained IoT

In the past few years, billions of Internet of Things (IoT) devices that lacked adequate security procedures were created and deployed, and more of these devices are on the way as a result of the development of Beyond 5G technologies. Because of their susceptibility to malware, there is a pressing need for reliable methods that can identify infected IoT devices within networks. Precise and early identification of IoT malware is inevitable to achieve IoT security. Nevertheless, prevailing studies of IoT malware detection mostly support certain platforms, need complicated deep learning (DL) models to achieve efficiency, and are centrally trained on the device. The purpose of this study is to introduce a new Federated Learning (FL) Framework, which has been given the name FLC-NET, in order to train numerous distributed edge devices to identify malware cooperatively. After the malware binaries have been encoded into image representations using FLC-NET, a lightweight convolutional network known as LC-NET is introduced to model these malware patterns directly from the image data without any data engineering being required. Because of its lightweight design, LC-NET is suited for use in devices with limited resource availability. After that, sophisticated adversarial training will be offered on FLC-NET in order to collect defensive knowledge against adversarial samples from a variety of clients who will be participating. The FLC-NET is experimentally evaluated on the public malware dataset, and it is demonstrated efficient (Accuracy: 96.1%, f1-score: 95.5), effective, scalable, and resistant to adversarial attacks.

Faster and Lighter Meteorological Satellite Image Classification by a Lightweight Channel-Dilation-Concatenation Net

With the development of satellite photography, meteorologists are inclined to rely on methods for the automatic and efficient classification of weather images. However, many popular networks require numerous parameters and a lengthy inference time, making them unsuitable for real-time classification tasks. To solve these problems, a lightweight convolutional network termed the Channel-Dilation-Concatenation network (CDC-net) is constructed for meteorological satellite image classification. When extracting features, CDC-net utilizes depth-wise convolution rather than standard convolution. Additionally, a FeatureCopy operation was employed instead of a half-convolution operation. CDC-net extracts high-dimensional features and contains a local importance-based pooling layer, reducing the network's depth, the number of network parameters and inference time. Based on these techniques, the CDC-net achieves an accuracy of 93.56% on the Large-Scale Satellite Cloud Image Database for Meteorological Research, with a GPU inference time of 3.261 ms and 1.12 million parameters. Because many weather images reveal multiple weather patterns, multiple labels are necessary. Therefore, we propose a prediction method and conduct experiments on multi-label data. Experiments on single-label and multi-label meteorological satellite image datasets demonstrate the superiority of the CDC-net over other structures. Thus, the proposed CDC-net can provide a faster and lighter solution in meteorological satellite image classification.

Rolling bearing fault diagnosis based on information fusion and parallel lightweight convolutional network

With the development of technologies such as Internet of Things and big data, the realization of fusion and cross analysis of multi-sensor signals provides the possibility for comprehensive condition monitoring and intelligent fault diagnosis of rolling bearings. Considering the bottleneck that deep neural networks with complex parameters and a single convolution kernel style may cause high computational effort and information loss, an intelligent fault diagnosis method based on information fusion and parallel lightweight convolutional network is proposed. Firstly, a normalized pulse energy kurtosis weighted rule is constructed to enable the fusion of multi-channel vibration signals. What’s more, a novel lightweight convolutional neural network is designed to achieve feature extraction and classification. Finally, a novel nonlinear piecewise activation function is introduced to further improve nonlinear learning ability. The effectiveness and superiority are verified by bearing data sets with different speeds and loads. Compared with other models, our diagnostic framework has better effect and provides a new idea for intelligent fault diagnosis.

MobileTrack: Siamese efficient mobile network for high‐speed UAV tracking

Recently, Siamese-based trackers have drawn amounts of attention in visual tracking field because of their excellent performance. However, visual object tracking on Unmanned Aerial Vehicles platform encounters difficulties under circumstances such as small objects and similar objects interference. Most existing tracking methods for aerial tracking adopt deeper networks or inefficient policies to promote performance, but most trackers can hardly meet real-time requirements on mobile platforms with limited computing resources. Thus, in this work, an efficient and lightweight siamese tracker (MobileTrack) is proposed for high-time Unmanned Aerial Vehicles tracking, realising the balance between performance and speed. Firstly, a lightweight convolutional network (D-MobileNet) is designed to enhance the characterisation ability of small objects. Secondly, an efficient object-aware module is proposed for local cross-channel information exchange, enhancing the feature information of the tracking object. Besides, an anchor-free region proposal network is introduced to predict the object pixel by pixel. Finally, deep and shallow feature information is fully utilised by cascading multiple anchor-free region proposal networks for accurate locating and robust tracking. Extensive experiments on the three Unmanned Aerial Vehicles benchmarks show that the proposed tracker achieves outstanding performance while keeping a beyond-real-time speed.

A Multi-Scale and Lightweight Bearing Fault Diagnosis Model with Small Samples

Currently, deep-learning-based methods have been widely used in fault diagnosis to improve the diagnosis efficiency and intelligence. However, most schemes require a great deal of labeled data and many iterations for training parameters. They suffer from low accuracy and over fitting under the few-shot scenario. In addition, a large number of parameters in the model consumes high computing resources, which is far from practical. In this paper, a multi-scale and lightweight Siamese network architecture is proposed for the fault diagnosis with few samples. The architecture proposed contains two main modules. The first part implements the feature vector extraction of sample pairs. It is composed of two lightweight convolutional networks with shared weights symmetrically. Multi-scale convolutional kernels and dimensionality reduction are used in these two symmetric networks to improve feature extraction and reduce the total number of model parameters. The second part takes charge of calculating the similarity of two feature vectors to achieve fault classification. The proposed network is validated by multiple datasets with different loads and speeds. The results show that the model has better accuracy, fewer model parameters and a scale compared to the baseline approach through our experiments. Furthermore, the model is also proven to have good generalization capability.

A DC Series Arc Fault Detection Method Based on a Lightweight Convolutional Neural Network Used in Photovoltaic System

Although photovoltaic (PV) systems play an essential role in distributed generation systems, they also suffer from serious safety concerns due to DC series arc faults. This paper proposes a lightweight convolutional neural network-based method for detecting DC series arc fault in PV systems to solve this issue. An experimental platform according to UL1699B is built, and current data ranging from 3 A to 25 A is collected. Moreover, test conditions, including PV inverter startup and irradiance mutation, are also considered to evaluate the robustness of the proposed method. Before fault detection, the current data is preprocessed with power spectrum estimation. The lightweight convolutional neural network has a lower computational burden for its fewer parameters, which can be ready for embedded microprocessor-based edge applications. Compared to similar lightweight convolutional network models such as Efficientnet-B0, B2, and B3, the Efficientnet-B1 model shows the highest accuracy of 96.16% for arc fault detection. Furthermore, an attention mechanism is combined with the Efficientnet-B1 to make the algorithm more focused on arc features, which can help the algorithm reduce unnecessary computation. The test results show that the detection accuracy of the proposed method can be up to 98.81% under all test conditions, which is higher than that of general networks.

A lightweight convolutional network for infrared object detection and tracking

As an important application of computer vision, visual tracking has being a fundamental topic. Compared with visible image, infrared image has the characteristic of low resolution, blurred contour and single color feature. Thus, it is still a challenge for infrared object tracking. Further, it is difficult to balance the real-time performance and accuracy. This paper proposed a method for target detection and tracking, with a deeper and lightweight MobileNet V2 structure as the backbone network. In the end, the tracker is tested on various datasets. Result shows that the tracker can get a balance between tracking accuracy and inference speed, which is crucial for deployment on mobile devices.

Facial Expression Recognition Based on Deep Binary Convolutional Network

Facial expression recognition (FER) is a challenging task because of the large number of parameters, low speed and insufficient feature representation of the expression region. In order to address the above-mentioned challenges, a deep binary convolutional network for FER is proposed. Firstly, a lightweight convolution network (BRNet) with parallel operation of binary convolution and traditional convolution is designed to reduce the complexity of network and improve the speed of recognition. Secondly, a dynamic radius strategy is constructed to extract binary features and form the attention weight of expression region so that the local features and global features can be fused effectively. Finally, the cross-entropy and <italic>L</italic>₂ loss are designed for quick and accurate expression classification. Experiments show that the average accuracy of the proposed method is 99.25% and 93.85% on CK+ and Oulu-CASIA respectively, which are higher than the other similar lightweight convolutional networks. The parameters and computation cost of network are 5.0×10⁵ bytes and 2.1×10⁶ bytes. In contrast, the computation cost of EfficientFace is about 77 times than proposed method. As a result, the effectiveness and lightness of proposed method have been proved.

Ship Detection Algorithm for SAR Images Based on Lightweight Convolutional Network

Ship Detection Algorithm for SAR Images Based on Lightweight Convolutional Network

Deep Learning-Based Emotion Detection

In order to make artificial intelligence smarter by detecting user emotions, this project analyzes and determines the current type of human emotions through computer vision, semantic recognition and audio feature classification. In facial expression recognition, for the problems of large number of parameters and poor real-time performance of expression recognition methods based on deep learning, Wang Weimin and Tang Yang Z. et al. proposed a face expression recognition method based on multilayer feature fusion with light-weight convolutional networks, which uses an improved inverted residual network as the basic unit to build a lightweight convolutional network model. Based on this method, this experiment optimizes the traditional CNN MobileNet model and finally constructs a new model framework ms_model_M, which has about 5% of the number of parameters of the traditional CNN MobileNet model. ms_model_M is tested on two commonly used real expression datasets, FER-2013 and AffectNet, the accuracy of ms_model_M is 74.35% and 56.67%, respectively, and the accuracy of the traditional MovbliNet model is 74.11% and 56.48% in the tests of these two datasets. This network structure well balances the recognition accuracy and recognition speed of the model. For semantic emotion detection and audio emotion detection, the existing models and APIs are used in this experiment.