A Model for Surface Defect Detection of Industrial Products Based on Attention Augmentation

<table class="table-group" id="tab2"><tr><td><table class="table"><tr><td class="thead-hr" colspan="3"><hr/></td></tr><tr class="thead"><td class="align_left">Number</td><td class="align_center">Methods</td><td class="align_center">mAP</td></tr><tr><td class="thead-hr" colspan="3"><hr/></td></tr><tr><td class="align_left">A</td><td class="align_center">CSPDarkNet53 + SPP + PANet</td><td class="align_center">95.97</td></tr><tr><td class="align_left">B</td><td class="align_center">A + 832 × 832 resolution</td><td class="align_center">95.33 (↓0.64)</td></tr><tr><td class="align_left">C</td><td class="align_center">A + 104 × 104 feature to P3</td><td class="align_center">95.75(↓0.22)</td></tr><tr><td class="align_left">D</td><td class="align_center">A + FDS block</td><td class="align_center">96.97 (↑1.00)</td></tr><tr><td class="align_left">E</td><td class="align_center">A + attention block</td><td class="align_center">97.42 (↑1.45)</td></tr><tr><td class="align_left">F</td><td class="align_center">Our method (D + E)</td><td class="align_center">98.52 (↑2.55)</td></tr><tr class="table-tr"><td colspan="3"><hr class="tbody-hr"/></td></tr></table></td></tr></table>

<div>Ablation studies on the internal dataset.</div>

Computational Intelligence and Neuroscience

tab2

Table 2

Table 2: A Model for Surface Defect Detection of Industrial Products Based on Attention Augmentation 