High-resolution remote sensing image target detection method of M-F-Y type lightweight convolutional neural network

Abstract

The invention discloses a high-resolution remote sensing image target detection method of an M-F-Y type lightweight convolutional neural network, and belongs to the field of remote sensing. The high-resolution remote sensing image target detection method comprises the following steps: firstly, constructing a feature pyramid network structure FPN on the basis of a lightweight convolutional neural network (CNN) model MobileNetV3-Small; extracting a high-resolution remote sensing image, fusing multi-scale depth features, and constructing an M-F-Y type lightweight convolutional neural network by jointly utilizing a YOLOv3tiny target detection framework; then, by constructing a complementary attention network structure, improving the attention to spatial position information of the target whileinhibiting a complex background; and finally, using a filter grafting strategy training model based on transfer learning to realize high-resolution remote sensing image target detection. The high-resolution remote sensing image target detection method can improve the target detection accuracy of the high-resolution remote sensing image while reducing the constraint on the high-speed computing power of the platform through less parameter quantity and lower delay, and can provide technical accumulation for the practicability of the target detection of the high-resolution remote sensing image.

Description

technical field [0001] An M-F-Y lightweight convolutional neural network object detection method for high-resolution remote sensing images belongs to the field of remote sensing. Background technique [0002] With the rapid development of remote sensing technology, the number of remote sensing images has increased dramatically, especially the mature application of high-resolution satellites such as IKONOS, Quickbird, WorldView, and GF-1, making the resolution of remote sensing images reach meter level. High-resolution remote sensing images contain rich spatial and texture features, as well as more complex spatial layout and geometric structure. Target detection on high-resolution remote sensing images is the basic work of remote sensing image interpretation. However, in the face of complex background interference and high-resolution remote sensing images with diverse ground structures, how to accurately and quickly detect objects has become one of the most important researc...

AI Technical Summary

Problems solved by technology

Third, the existing labeled high-resolution remote sensing image target detection data set has a limited sample size, which will cause over-fitting problems when used for training networks, resulting in poor robustness of model learning features and poor model generalization ability; at the same time CNN often contains a large number of useless convolution kernels during training, which hardly helps the performance of the model while occupying storage space and computing space

[0005] As we all know, in the development of CNN, in order to pursue accuracy, CNN often becomes deeper and deeper. From VGGNet to ResNet, the number of parameters is increasing, resulting in poor real-time performance of the model and high computational complexity.

Images