Small sample target detection method and system based on support and query samples

Abstract

The invention discloses a small sample target detection method and system based on support and query samples, comprising support sample and query sample feature extraction, support sample weighting based on query sample guidance, query feature enhancement based on support sample guidance, scoring and screening of candidate boxes, and mixed loss function calculation. A small sample learning mechanism is introduced into a deep target detection framework, and a set of small sample target detection system with high accuracy is established. The method is simple in framework, convenient to use, high in expandability and high in interpretability, and results of small sample target detection of two mainstream visual attribute data sets exceed those of an existing method. The method can provide basic framework and algorithm support for the target detection technology in the military and industrial application field, and can be easily expanded to other small sample learning tasks.

Description

technical field [0001] The invention relates to the technical field of machine learning, in particular to a small sample target detection method and system based on support and query samples. Background technique [0002] Target detection technology is a basic task in computer vision tasks, which aims to locate and classify target category objects from images. Target detection technology has a wide range of applications, and it provides basic support for some downstream tasks, such as instance segmentation, scene understanding, pose estimation and other tasks. Existing deep object detection models have achieved good accuracy in some categories, but rely heavily on large-scale calibrated datasets. However, in real scenarios, there are problems such as unbalanced distribution of data samples and unsupervised samples. Therefore, how to perform effective target detection in the case of insufficient sample size has become an open problem in the field of computer vision. Small-s...

AI Technical Summary

Problems solved by technology

[0004] 1) The accuracy of the method based on the modification of the one-stage target detection framework is often low

[0005] 2) The method based on the modification of the two-stage target detection framework often does not make full use of the information of the supporting samples to guide the generation of candidate boxes in the first stage and the screening of candidate boxes in the second stage

[0006] 3) Under K-shot, all support samples are treated equally, without considering the contribution of different support samples to the current query sample

[0007] 4) Overall, the accuracy of existing methods is low, and it is difficult to reach the level of actual use

However, it takes a lot of time to input products by scanning, and the settlement efficiency is low, which greatly reduces the user experience

Images