Image segmentation method and system based on wide-residual pyramid pooling network

Abstract

The invention discloses an image segmentation method and system based on a wide-residual pyramid pooling network. The method comprises: inputting a to-be-segmented image; standardizing the to-be-segmented image; acquiring a WRN-PPNet model; carrying out pretreatment on a training image and increasing a mode and number of training images by using a data expansion method to obtain a training image set; carrying out model training based on the WRN-PPNet model and training image set to generate a WRN-PPNet segmentation model; and according to the to-be-segmented image, obtaining an image segmentation result by the WRN-PPNet segmentation model. According to the method, the image can be segmented automatically based on the WRN-PPNet to realize target object segmentation without the restriction of the type of the to-be-segmented image; the adaptability is high and the model performance is good; and the accuracy and convenience of image segmentation are improved effectively.

Description

technical field [0001] The invention relates to the technical fields of image processing and computer vision, and particularly designs an image segmentation method and system based on a wide residual pyramid pooling network. Background technique [0002] In related technologies, FCN (fully convolutional network, fully convolutional network) has opened the door to deep learning for image semantic segmentation. Since then, most deep learning models for image semantic segmentation have made some improvements based on FCN. FCN uses the existing CNNs (convolutional neural networks, convolutional network) as a visual model, learns hierarchical features, and then changes the last fully connected layer of the classification network to a fully convolutional layer, and then outputs a feature map to replace the classification score. Finally, deconvolution is performed on these feature maps to produce dense pixel-level labeled output maps. This network model realizes the end-to-end sol...

AI Technical Summary

Problems solved by technology

[0003] However, the models mentioned above are all aimed at specific semantic segmentation problems, and have been improved to varying degrees on the basis of FCN, and none of the models can be well used to solve more and different image semantic segmentation problems. Therefore, there is still a lot of room for exploration in the deep learning network model structure for image semantic segmentation.

In related technologies, the traditional segmentation method is to extract appropriate features, and then segment according to the features, but this type of method can only extract shallow features, and the operation is complicated, the applicability is not strong, and it is difficult to popularize

Images