Figure 3 | Scientific Reports

Figure 3

From: BD-YOLOv8s: enhancing bridge defect detection with multidimensional attention and precision reconstruction

Figure 3

Illustration of multiplying four types of attentions in ODConv to convolutional kernels progressively. (a) ODConv Convolution attention operations along spatial dimensions, (b) ODConv Convolution attention operations along the in-put channel dimension, (c) ODConv filters convolution attention operations along the output channel dimension, (d) ODConv convolutions attention operations along the spatial dimension of the convolution kernel kerne.

Back to article page