Yue Liu

Accurate and efficient PCB defect detection is a key component of automated quality inspection in electronic manufacturing. In recent years, with the rapid development of deep learning, surface defect detection technology has made significant progress. However, the small size, diverse morphology, and complex background of PCB defects still pose significant challenges for accurate detection. To address these issues, this paper proposes a PCB surface defect detection network (MPHF-Net) based on transformer multi-dimensional perception and high-low frequency fusion. First, this paper proposes a cross stage partial multi-scale edge enhancement module to enhance multi-scale feature extraction capabilities and improve the representation of defect edge details. Second, we propose a lightweight oriented spatial convolution method to capture directional defect features. Finally, to enhance information interaction between features at different scales, we introduce the CGAFusion feature fusion mechanism to effectively integrate high-frequency texture information with low-frequency semantic information. The experimental results show that the network proposed in this paper exhibits excellent performance on PKU-Market-PCB and DeepPCB datasets. Compared with the baseline model, mAP50 and mAP50–95 improved by 2.9% and 1.2% respectively, reduced computational complexity by 11.1%, and reduced parameter count by 26.6%. The overall performance is better than popular object detection networks. The results validated the detection performance of MPHF-Net, providing a new approach for industrial quality inspection.

In previous research on medical image classification, CNNs cannot capture long-range dependencies with local receptive fields and result in poor classification performance. Transformer-based models are limited by the quadratic computational complexity of the self-attention mechanism, especially when processing high-resolution medical images. It is difficult to deploy them in limited computational settings without sacrificing performance. Mamba-based models have attracted a lot of interests in computer vision due to their linear computation complexity. Despite their low FLOPs, Mamba-based models with less parameters perform sub-optimally in image classification tasks. To overcome the limitations of Mamba-based models, we propose InceptionMamba, a model that combines lightweight design with high accuracy for medical image classification tasks.Inspired by the impressive performance of the Inception architecture at a relatively low computational cost, we introduce Inception modules Mamba-based model. Meanwhile, a channel attention mechanism is employed to improve performance. Additionally, we conduct an in-depth analysis of the modeling capabilities of State Space Model (SSM) from the perspective of frequency response, revealing that it is better suited for medical images dominated by low-frequency components rather than natural images dominated by high-frequency information. InceptionMamba demonstrates competitive performance on medical image classification tasks, surpassing most state-of-the-art methods. The source code is publicly available at https://github.com/pepper1329/InceptionMamba.

The challenge of semantic segmentation with scarce pixel-level annotations has induced many self-supervised works, however most of which essentially train an image encoder or a segmentation head that produces finer dense representations, and when performing segmentation inference they need to resort to supervised linear classifiers or traditional clustering. Segmentation by dataset-level clustering not only deviates the real-time and end-to-end inference practice, but also escalates the problem from segmenting per image to clustering all pixels at once, which results in downgraded performance. To remedy this issue, we propose a novel self-supervised semantic segmentation training and inferring paradigm where inferring is performed in an end-to-end manner. Specifically, based on our observations in probing dense representation by image-level self-supervised ViT, i.e. semantic inconsistency between patches and poor semantic quality in non-salient regions, we propose prototype-image alignment and global-local alignment with attention map constraint to train a tailored Transformer Decoder with learnable prototypes and utilize adaptive prototypes for segmentation inference per image. Extensive experiments under fully unsupervised semantic segmentation settings demonstrate the superior performance and the generalizability of our proposed method.

Octane number is a measure of gasoline’s ability to resist detonation and combustion in the cylinder; the higher the value, the better the resistance to detonation. The accurate prediction of octane loss during gasoline refining could facilitate production management and ensure gasoline octane. The backpropagation neural network is a traditional method adopted for the octane loss prediction, but there exists the issues of low training accuracy and poor generalization in the traditional BP neural network model caused by randomly generated weights and thresholds at input. In this paper, we propose a novel approach to optimize the weights and thresholds for gasoline octane number prediction based on a self-adaptive genetic algorithm. The experimental result shows that the proposed model outperforms in accuracy and generalization in the competition with the traditional BP neural network. The coefficient of determination of the performance index in the experiment is improved from 0.81502 to 0.95628, and the average prediction error among 10 groups of experiments was reduced from 0.0061 to 0.0041.

In the era of digital manufacturing, huge amount of image data generated by manufacturing systems cannot be instantly handled to obtain valuable information due to the limitations (e.g., time) of traditional techniques of image processing. In this paper, we propose a novel self-supervised self-attention learning framework—TriLFrame for image representation learning. The TriLFrame is based on the hybrid architecture of Convolutional Network and Transformer. Experiments show that TriLFrame outperforms state-of-the-art self-supervised methods on the ImageNet dataset and achieves competitive performances when transferring learned features on ImageNet to other classification tasks. Moreover, TriLFrame verifies the proposed hybrid architecture, which combines the powerful local convolutional operation and the long-range nonlocal self-attention operation and works effectively in image representation learning tasks.