An Improved Ship Classification Method Based on YOLOv7 Model with Attention Mechanism

Algorithm 1

Training of YOLOv7-RCBAM.

Input: Training set and ground truths

Output: Trained model

Initialization: Learning rate , batch size , and set each parameter value epoch, weights.

For to do

 //YOLOv7-RCBAM

 Preprocess by mosaic: = ImagePreprocessing()

 Extract feature information in backbone: Features = backbone()

 Fuse features by feature fusion module: Fused = FeatureFusion(features)

 Attention module extract feature: Attention = RCBAM(fused)

 Objectness, anchor_shape, feature_cls, feature_loc = YOLOHead(attention)

 //Calculate loss and gradient descent

 Calculate gradients: Weight = backpropagation()

End: save weights of model