Visualization for Explanation of Deep Learning-Based Defect Detection Model Using Class Activation Map

Recently,convolutional neural network(CNN)-based visual inspec-tion has been developed to detect defects on building surfaces automatically.The CNN model demonstrates remarkable accuracy in image data analysis;however,the predicted results have uncertainty in providing accurate informa-tion to users because of the“black box”problem in the deep learning model.Therefore,this study proposes a visual explanation method to overcome the uncertainty limitation of CNN-based defect identification.The visual repre-sentative gradient-weights class activation mapping(Grad-CAM)method is adopted to provide visually explainable information.A visualizing evaluation index is proposed to quantitatively analyze visual representations;this index reflects a rough estimate of the concordance rate between the visualized heat map and intended defects.In addition,an ablation study,adopting three-branch combinations with the VGG16,is implemented to identify perfor-mance variations by visualizing predicted results.Experiments reveal that the proposed model,combined with hybrid pooling,batch normalization,and multi-attention modules,achieves the best performance with an accuracy of 97.77%,corresponding to an improvement of 2.49%compared with the baseline model.Consequently,this study demonstrates that reliable results from an automatic defect classification model can be provided to an inspector through the visual representation of the predicted results using CNN models.

Density-dependent carrier-envelope phase shift in attosecond pulse generation from relativistically oscillating mirrors

The carrier-envelope phase(CEP)φ_(0)is one of the key parameters in the generation of isolated attosecond pulses.In particular,“cosine”pulses(φ_(0)=0)are best suited for generation of single attosecond pulses in atomic media.Such“cosine”pulses have the peak of the most intense cycle aligned with the peak of the pulse envelope,and therefore have the highest contrast between the peak intensity and the neighboring cycles.In this paper,the dynamics of single attosecond pulse generation from a relativistically oscillating plasma mirror is investigated.We use an elementary analytical model as well as particle-in-cell simulations to study few-cycle attosecond pulses.We find that the phase of the field driving the surface oscillations depends on the plasma density and preplasma scale length.This leads us to a counterintuitive conclusion:for the case of normal incidence and a sharp plasma-vacuum boundary,the CEP required for the generation of a single attosecond pulse phase is closer toφ_(0)=π/2(a“sine”pulse),with the exact value depending on the plasma parameters.