An intelligent detection method for power line faults based on sk-yolov3

Abstract

The invention discloses a power line fault intelligent detection method based on SK-YOLOv3, comprising the following steps: 1) collecting a power line fault data set; 2) improving the score of a feature map; 3) generating a prediction frame. This method improves the detection accuracy and makes the detection more detailed.

Description

technical field [0001] The invention belongs to the field of computer vision and image detection, and relates to a power line fault intelligent detection method, in particular to a power line fault intelligent detection method based on SK-YOLOv3. Background technique [0002] Fault detection is one of the important applications of machine vision in the field of industrial manufacturing. It can improve factory production efficiency and reduce human labor, and can monitor product quality in real time. How to accurately detect power line faults has become a difficult point in current research. At present, there are two main research methods. One is to use traditional image recognition methods to extract and classify image features, and the other is to directly use neural networks to identify fault categories. [0003] Traditional image detection algorithms are mainly based on threshold segmentation, spectral methods, and grayscale histograms. Commonly used edge detection opera...

AI Technical Summary

Problems solved by technology

Such as SSD[[6]Liu W,AnguelovD,Erhan D,et al.SSD:single shot multibox detector[C] / / Proc of EuropeanConference on ComputerVision. Amsterdam,Nederland:Springer,2016:21-37.], YOLO[ [7]Redmon J, Divvala S, Girshick R, et al.You only look once:unified,real timeobject detection[C] / / Computer Vision and Pattern Recognition.Las Vegas,USA:IEEE,2015:779-788] series, etc. , the above network has high accuracy for detecting large objects, but for smaller objects, there are problems such as false detection and missed detection.

[0004] In fact, the detection of small faults has always had shortcomings such as small target detection field of view, single aspect ratio of the detection image, and low detection accuracy.

In order to solve the above problems, many scholars have enhanced the network performance by improving the structure, but there are shortcomings of huge memory usage and low detection speed.

Images