Concrete crack segmentation method and system based on deep learning and attention mechanism

The invention provides a concrete crack segmentation method and system based on deep learning and an attention mechanism, and the method comprises the following steps: inputting an original image into a classical convolutional neural network ResNet101, and obtaining a feature map through a convoluti...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: HANG JIAQI, LI YANCHENG, LAI TAO
Format: Patent
Sprache:chi ; eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The invention provides a concrete crack segmentation method and system based on deep learning and an attention mechanism, and the method comprises the following steps: inputting an original image into a classical convolutional neural network ResNet101, and obtaining a feature map through a convolutional layer, a pooling layer, a batch standardization layer and an activation function; highlighting crack features in the feature map by using a self-attention mechanism VH-CAM; the low-level feature map is guided by a channel attention mechanism ECAUM and then is subjected to feature fusion with the high-level feature map subjected to up-sampling, and a feature map with the same size as the original image is obtained; softmax prediction is carried out on the feature map with the same size as the original image, the category of each pixel in the image is obtained, and pixel-level segmentation of the crack is achieved. According to the invention, high-precision crack segmentation can be realized, so that the method