An Image-Based Deep Learning Approach with Improved DETR for Power Line Insulator Defect Detection

Figure 1

The model structure of Transformer [15]. The three arrows generated from one are part of the self-attention mechanism, which corresponds to , , and in Equation (1) generated from the same input by three different linear projections. That is why it is called “Self-Attention.” The two arrows making from one are the and generated from the encoder and along with the from the previous layers of decoder as inputs to the multi-head attention which is also called “Cross-Attention.”