Search

Salient object detection aims to identify the most attentive regions of an image, and as an image processing technology, it is widely used in imaging systems of many consumer electronics. With the development of deep learning, a series of notable results has achieved in salient object detection field. However, there are two intractable challenges still remain to be studied: 1) the issue of model lightweight; 2) model robustness. To address these two problems, we propose a feature filtering and fusion module based on a novel selection mechanism (SM), which can adaptively discover and fuse multi-scale features. Compared to other methods, it reduces the number of features that need to be processed, greatly reducing the complexity of the salient object detection model. We simultaneously explored the role of global contextual information and global multi-scale features in saliency detection task and validated them. Based on the SM block, we propose a lightweight salient object detection network named SMNet. Extensive experimental results on six datasets show that our model achieves satisfactory results, which can greatly save storage space and improve data transmission speed of image acquisition equipment.

Conducting surface defect detection on lead frames is a crucial operation in semiconductor manufacturing that significantly impacts product quality. This study develops a lead frame dataset and introduces a novel network tailored to detect minute target defects on lead frame surfaces across the production process. Initially, this study proposes a learnable hierarchical graph network (LHGNet) for effective feature extraction of small target defects, using learnable dynamic convolution modules for targeted feature extraction. Subsequently, this study introduces multiattention-based high-level screening-feature pyramid networks (MHS-FPN) and mixed cascade grouped attention (MCGA) modules, which enhance the feature expression of small targets while improving model classification performance. In addition, a loss function is specifically tailored for the lead frame scenario. Using the lead frame dataset and comparing it with competing methods, the feasibility and effectiveness of this model in detecting surface defects on lead frames in production lines are validated. The proposed surface defect detection you only look once (SDD-YOLO) network achieves a mean average precision (mAP) of 87.9% and an overall missed detection rate (OMD) of 2.06%, demonstrating its capability to either replace or assist manual inspections in assessing product quality.

The fast and accurate recognition of the container code and type is very important to speed up the passage of trucks through the gate and improve the efficiency of port. In this article, a practical unified network (PUN) is proposed for the first time to recognize arbitrary-oriented container code and type. The model is based on the encoder-decoder network, which integrates the detection of the second part of container code, the semantic segmentation of the three parts of container code and type, and the classification of single character features of container code and type. The ablation experiments verified the performance of each part of PUN and obtained a stable and reliable model. The comparison experiments compare PUN with the state-of-the-art text detection and end-to-end text recognition algorithms, and the results show that PUN achieves almost the best performance in both container code and type detection and end-to-end recognition. The proposed algorithm has only 15.53 M parameters, with a frame per second (FPS) of 4.86 with an input size of $768\times 768$ , which meets the company’s needs. Our model has been deployed and running in multismart gates of Shanghai Port for nearly a year, showed certain practical application capabilities.

Emergency prediction and driver attention prediction are fundamental tasks within the realm of self-driving vehicles and assistant driving systems. The utilization of visual saliency detection in these tasks has garnered considerable attention, owing to its inherent advantages. However, current research on traffic saliency detection primarily focuses on emulating the human visual system for attention allocation in traffic scenes, neglecting the detection of complete salient objects. In this paper, we propose the Traffic Salient Object Detection Using a Feature Deep Interaction and Guidance Fusion Network (TFGNet). Different from previous methods, our method detects the complete objects that attract human attention in natural traffic scenes, rather than a certain point without object semantic information, which can provide assistance for target recognition tasks in the domain of intelligent driving. Moreover, we propose a traffic salient object detection framework based on feature interaction and guidance fusion, enabling the detection of salient objects across varying scales. Experimental results on multiple benchmark datasets demonstrate that, compared to the state-of-the-art methods, our method exhibits superior performance in terms of precision, recall, and error rate.

In this article, an improved smoothed particle magnetohydrodynamics (SPMHDs) method is proposed to study micro-explosion phenomena of cathode micro-protrusion under high-density electron emission. First, the improved SPMHD method is established by introducing the coupling effect of the heat conduction effect and Joule heating effect into the basic SPMHD method. Furthermore, the latent heat of phase transition is processed and the equations of state for different phase states are introduced into the basic SPMHD method. Then, the dynamic characteristics of the micro-protrusion during the micro-explosion, such as the temperature rise, the phase transition, the surface destruction, and the diffusion of the metal vapor, are quantitatively obtained and discussed based on the improved SPMHD method. The results show that the high-density electron emission could cause the significant temperature increase of the micro-protrusion and hence a micro-explosion. The action of the pressure exerted by the micro-explosion could lead to the formation of a crater on the cathode surface. The metal vapor produced by the micro-explosion would propagate into the surrounding vacuum gap at a high velocity, which may lead to the failure of the vacuum insulation. The results of this article may provide some useful information to further understand the mechanism of vacuum breakdown initiated by the electron emission.