Search Results (10)

Space-time image velocimetry (STIV) plays an important role in river velocity measurement due to its safety and efficiency. However, its practical application is affected by complex scene conditions, resulting in significant errors in the accurate estimation of texture angles. This paper proposes a method to predict the texture angles in frequency domain images based on an improved ShuffleNetV2. The second 1 × 1 convolution in the main branch of the downsampling unit and basic unit is deleted, the kernel size of the depthwise separable convolution is adjusted, and a Bottleneck Attention Module (BAM) is introduced to enhance the ability of capturing important feature information, effectively improving the precision of texture angles. In addition, the measured data from a current meter are used as the standard for comparison with established and novel approaches, and this study further validates its methodology through comparative experiments conducted in both artificial and natural river channels. The experimental results at the Agu, Panxi, and Mengxing hydrological stations demonstrate that the relative errors of the discharge measured by the proposed method are 2.20%, 3.40%, and 2.37%, and the relative errors of the mean velocity are 1.47%, 3.64%, and 1.87%, which affirms it has higher measurement accuracy and stability compared with other methods. Full article

Wheat is one of the most essential food crops globally, but diseases significantly threaten its yield and quality, resulting in considerable economic losses. The identification of wheat diseases faces challenges, such as interference from complex environments in the field, the inefficiency of traditional machine learning methods, and difficulty in deploying the existing deep learning models. To address these challenges, this study proposes a multi-scale feature fusion shuffle network model (MFFSNet) for wheat disease identification from complex environments in the field. MFFSNet incorporates a multi-scale feature extraction and fusion module (MFEF), utilizing inflated convolution to efficiently capture diverse features, and its main constituent units are improved by ShuffleNetV2 units. A dual-branch shuffle attention mechanism (DSA) is also integrated to enhance the model’s focus on critical features, reducing interference from complex backgrounds. The model is characterized by its smaller size and fast operation speed. The experimental results demonstrate that the proposed DSA attention mechanism outperforms the best-performing Squeeze-and-Excitation (SE) block by approximately 1% in accuracy, with the final model achieving 97.38% accuracy and 97.96% recall on the test set, which are higher than classical models such as GoogleNet, MobileNetV3, and Swin Transformer. In addition, the number of parameters of this model is only 0.45 M, one-third that of MobileNetV3 Small, which is very suitable for deploying on devices with limited memory resources, demonstrating great potential for practical applications in agricultural production. Full article

To better utilize multimodal information for agriculture applications, this paper proposes a cherry tomato bunch detection network using dual-channel cross-feature fusion. It aims to improve detection performance by employing the complementary information of color and depth images. Using the existing YOLOv8_n as the baseline framework, it incorporates a dual-channel cross-fusion attention mechanism for multimodal feature extraction and fusion. In the backbone network, a ShuffleNetV2 unit is adopted to optimize the efficiency of initial feature extraction. During the feature fusion stage, two modules are introduced by using re-parameterization, dynamic weighting, and efficient concatenation to strengthen the representation of multimodal information. Meanwhile, the CBAM mechanism is integrated at different feature extraction stages, combined with the improved SPPF_CBAM module, to effectively enhance the focus and representation of critical features. Experimental results using a dataset obtained from a commercial greenhouse demonstrate that DCFA-YOLO excels in cherry tomato bunch detection, achieving an mAP50 of 96.5%, a significant improvement over the baseline model, while drastically reducing computational complexity. Furthermore, comparisons with other state-of-the-art YOLO and other object detection models validate its detection performance. This provides an efficient solution for multimodal fusion for real-time fruit detection in the context of robotic harvesting, running at 52fps on a regular computer. Full article

Running machine learning algorithms for image classification locally on small, cheap, and low-power microcontroller units (MCUs) has advantages in terms of bandwidth, inference time, energy, reliability, and privacy for different applications. Therefore, TinyML focuses on deploying neural networks on MCUs with random access memory sizes between 2 KB and 512 KB and read-only memory storage capacities between 32 KB and 2 MB. Models designed for high-end devices are usually ported to MCUs using model scaling factors provided by the model architecture’s designers. However, our analysis shows that this naive approach of substantially scaling down convolutional neural networks (CNNs) for image classification using such default scaling factors results in suboptimal performance. Consequently, in this paper we present a systematic strategy for efficiently scaling down CNN model architectures to run on MCUs. Moreover, we present our CNN Analyzer, a dashboard-based tool for determining optimal CNN model architecture scaling factors for the downscaling strategy by gaining layer-wise insights into the model architecture scaling factors that drive model size, peak memory, and inference time. Using our strategy, we were able to introduce additional new model architecture scaling factors for MobileNet v1, MobileNet v2, MobileNet v3, and ShuffleNet v2 and to optimize these model architectures. Our best model variation outperforms the MobileNet v1 version provided in the MLPerf Tiny Benchmark on the Visual Wake Words image classification task, reducing the model size by 20.5% while increasing the accuracy by 4.0%. Full article

The driving operations of the subway system are of great significance in ensuring the safety of trains. There are several hand actions defined in the driving instructions that the driver must strictly execute while operating the train. The actions directly indicate whether equipment is normally operating. Therefore, it is important to automatically sense the region of the driver and detect the actions of the driver from surveillance cameras to determine whether they are carrying out the corresponding actions correctly or not. In this paper, a lightweight two-stage model for subway driver action sensoring and detection is proposed, consisting of a driver detection network to sense the region of the driver and an action recognition network to recognize the category of an action. The driver detection network adopts the pretrained MobileNetV2-SSDLite. The action recognition network employs an improved ShuffleNetV2, which incorporates a spatial enhanced module (SEM), improved shuffle units (ISUs), and shuffle attention modules (SAMs). SEM is used to enhance the feature maps after convolutional downsampling. ISU introduces a new branch to expand the receptive field of the network. SAM enables the model to focus on important channels and key spatial locations. Experimental results show that the proposed model outperforms 3D MobileNetV1, 3D MobileNetV3, SlowFast, SlowOnly, and SE-STAD models. Furthermore, a subway driver action sensoring and detection system based on a surveillance camera is built, which is composed of a video-reading module, main operation module, and result-displaying module. The system can perform action sensoring and detection from surveillance cameras directly. According to the runtime analysis, the system meets the requirements for real-time detection. Full article

Due to complex environmental factors such as illumination, shading between leaves and fruits, shading between fruits, and so on, it is a challenging task to quickly identify red jujubes and count red jujubes in orchards. A counting method of red jujube based on improved YOLOv5s was proposed, which realized the fast and accurate detection of red jujubes and reduced the model scale and estimation error. ShuffleNet V2 was used as the backbone of the model to improve model detection ability and light the weight. In addition, the Stem, a novel data loading module, was proposed to prevent the loss of information due to the change in feature map size. PANet was replaced by BiFPN to enhance the model feature fusion capability and improve the model accuracy. Finally, the improved YOLOv5s detection model was used to count red jujubes. The experimental results showed that the overall performance of the improved model was better than that of YOLOv5s. Compared with the YOLOv5s, the improved model was 6.25% and 8.33% of the original network in terms of the number of model parameters and model size, and the Precision, Recall, F1-score, AP, and Fps were improved by 4.3%, 2.0%, 3.1%, 0.6%, and 3.6%, respectively. In addition, RMSE and MAPE decreased by 20.87% and 5.18%, respectively. Therefore, the improved model has advantages in memory occupation and recognition accuracy, and the method provides a basis for the estimation of red jujube yield by vision. Full article

Assessing the quality of agricultural products is an essential step to reduce food waste. The problems of overly complex models, difficult to deploy to mobile devices, and slow real-time detection in the application of deep learning in agricultural product quality assessment requiring solutions. This paper proposes a lightweight method based on ShuffleNetV2 to identify phenotypic diseases in corn seeds and conduct experiments on a corn seed dataset. Firstly, Cycle-Consistent Adversarial Networks are used to solve the problem of unbalanced datasets, while the Efficient Channel Attention module is added to enhance network performance. After this, a $7\times 7$ depthwise convolution is used to increase the effective receptive field of the network. The repetitions of basic units in ShuffleNetV2 are also reduced to lighten the network structure. Finally, experimental results indicate that the number of model parameters are 0.913 M, the computational volume is 44.75 MFLOPs and 88.5 MMAdd, and the recognition accuracy is 96.28%. The inference speed of about 9.71 ms for each image was tested on a mobile portable laptop with only a single CPU, which provides a reference for mobile deployment. Full article

Owing to their high accuracy, deep convolutional neural networks (CNNs) are extensively used. However, they are characterized by high complexity. Real-time performance and acceleration are required in current CNN systems. A graphics processing unit (GPU) is one possible solution to improve real-time performance; however, its power consumption ratio is poor owing to high power consumption. By contrast, field-programmable gate arrays (FPGAs) have lower power consumption and flexible architecture, making them more suitable for CNN implementation. In this study, we propose a method that offers both the speed of CNNs and the power and parallelism of FPGAs. This solution relies on two primary acceleration techniques—parallel processing of layer resources and pipelining within specific layers. Moreover, a new method is introduced for exchanging domain requirements for speed and design time by implementing an automatic parallel hardware–software co-design CNN using the software-defined system-on-chip tool. We evaluated the proposed method using five networks—MobileNetV1, ShuffleNetV2, SqueezeNet, ResNet-50, and VGG-16—and FPGA processors—ZCU102. We experimentally demonstrated that our design has a higher speed-up than the conventional implementation method. The proposed method achieves 2.47×, 1.93×, and 2.16× speed-up on the ZCU102 for MobileNetV1, ShuffleNetV2, and SqueezeNet, respectively. Full article

High-order harmonics in the traction power supply show negative effects on the safe and stable operation of the railway transportation system. There is a fixed resonant frequency in the traction network. When the harmonic current frequency produced by the locomotive matches the resonant frequency of the traction network, it will cause high-frequency resonant overvoltage. The propagation path of the high-order harmonics of the traction load is analyzed based on a V/v wiring traction transformer. The propagation characteristics of high-order harmonics on self-used equipment at 380 V low-voltage side and 27.5 kV high-voltage side are expounded. A simulation model for the low-voltage self-consumption power system is established and the singular value decomposition algorithm is proposed to identify the harmonic impedance. The simulation results show that the proposed method can reduce the error to within 0.1%. Under realistic conditions, the overvoltage caused by high-order harmonics is difficult to identify. To solve this problem, an overvoltage identification algorithm for Electric Multiple Units based on a convolutional neural network is proposed. The ShuffleNet neural network model is then used to identify high-order harmonics overvoltage and other types of overvoltage. The overall accuracy of the proposed classification model can be improved from 97.12% to 98.44%. Better recognition and classification performances can also be achieved. Full article

The rolling bearing is an important part of the train’s running gear, and its operating state determines the safety during the running of the train. Therefore, it is important to monitor and diagnose the health status of rolling bearings. A convolutional neural network is widely used in the field of fault diagnosis because it does not require feature extraction. Considering that the size of the network model is large and the requirements for monitoring equipment are high. This study proposes a novel bearing fault diagnosis method based on lightweight network ShuffleNet V2 with batch normalization and L2 regularization. In the experiment, the one-dimensional time-domain signal is converted into a two-dimensional Time-Frequency Graph (TFG) using a short-time Fourier transform, though the principle of graphics to enhance the TFG dataset. The model mainly consists of two units, one for extracting features and one for spatial down-sampling. The building units are repeatedly stacked to construct the whole model. By comparing the proposed method with the origin ShuffleNet V2, machine learning model and state-of-the-art fault diagnosis model, the generalization of the proposed method for bearing fault diagnosis is verified. Full article