A Small Target Semantic Segmentation Method Combined with Target Detection

Abstract

The invention discloses a small target semantic segmentation method combined with target detection. The steps are: building a DeepLab-Attention semantic segmentation network, training the network to obtain an overall semantic segmentation model; making a small target detection data set and a small target semantic segmentation data set; The small target detection dataset trains the small target detection network based on YOLOv2; designs a small target semantic segmentation network, uses the small target semantic segmentation dataset to train the network, and obtains the small target semantic segmentation model; in the test phase, the test images are used as the above-mentioned The input of the overall semantic segmentation model and the small target detection network is to obtain the segmentation result of the entire image and the small target bounding box existing in the image, and correct it through the small target semantic segmentation model. The invention can greatly reduce the difficulty of segmenting the small target, thereby effectively improving the segmenting performance of the small target.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a small target semantic segmentation method combined with target detection. Background technique [0002] Image semantic segmentation is one of the three major tasks of computer vision. Its goal is to classify each pixel in the image and obtain a semantic segmentation map of the image. From the perspective of traditional image segmentation, image semantic segmentation is to segment the image into multiple regions at the semantic level, and then assign appropriate category labels to each region. At present, semantic segmentation has a wide range of applications in autonomous driving, real-time road monitoring, automatic virtual fitting, and medical disease systems. Before the rise of deep learning, the main method of semantic segmentation was to use the conditional random field model to build a probability graph model. In recent years, due to the strong learn...

AI Technical Summary

Problems solved by technology

However, the number of pixels contained in the area where the small target is located is much less than that contained in other targets. When the pixels in the small target area are misclassified, it will not have a great impact on the total loss. Therefore, Such a loss function is not suitable for segmentation of small objects

Images