SIA OpenIR  > 数字工厂研究室
基于YOLOv4-tiny的铁矿石品位识别技术研究
Alternative TitleResearch on Iron Ore Grade Identification Technology Based on YOLOv4-tiny
何文轩1; 荆洪迪2; 柳小波2; 于健洋1; 孙效玉1
Department数字工厂研究室
Source Publication金属矿山
ISSN1001-1250
2021
Issue10Pages:150-154
Contribution Rank2
Funding Organization国家自然科学基金项目(编号:51674061)
Keyword铁矿 矿石品位 图像识别 YOLOv4-tiny 特征识别
Abstract

为实现矿山现场对不同铁矿石品位的智能化识别以及模型在便携设备上的搭载,对不同铁矿石的不同品位进行了数据增强处理并选择YOLOv4-tiny作为训练的神经网络算法。YOLOv4-tiny深度学习神经网络框架,采用CSPdarknet53tiny作为主干提取网络并结合FPN对岩石图像进行特征提取和学习,在训练过程中采用迁移学习思想以及早停法对训练进行加速,进而生成铁矿石品位识别模型。最终通过测试集验证,模型对于每种矿石品位图像识别正确率大于91%。对于不同环境拍摄的图像以及视频识别也超过80%。模型可以很好地区分不同品位的铁矿石,试验证明模型的鲁棒性较强。

Other Abstract

In order to realize the intelligent recognition of different iron ore grades and the loading of models on portable devices,data enhancement processing is carried out on different grades of different iron ore and YOLOv4-tiny is selected as the neural network algorithm for training. YOLOv4-tiny deep learning neural network framework,using CSPdarknet53_tiny as the backbone extraction network and combined with FPN for feature extraction and learning of rock images. In the training process,migration learning ideas and early stopping methods are used to accelerate training to generate iron ore grades Identify the model. Finally,it is verified by the test set that the model has an image recognition accuracy rate of greater than 91% for each ore grade. The recognition of images and videos taken in different environments also exceeds 80%. The model can distinguish iron ore of different grades well,and the test proves that the model is robust.

Language中文
Document Type期刊论文
Identifierhttp://ir.sia.cn/handle/173321/29845
Collection数字工厂研究室
Corresponding Author何文轩
Affiliation1.东北大学智慧矿山研究中心
2.中国科学院沈阳自动化研究所数字工厂研究室
Recommended Citation
GB/T 7714
何文轩,荆洪迪,柳小波,等. 基于YOLOv4-tiny的铁矿石品位识别技术研究[J]. 金属矿山,2021(10):150-154.
APA 何文轩,荆洪迪,柳小波,于健洋,&孙效玉.(2021).基于YOLOv4-tiny的铁矿石品位识别技术研究.金属矿山(10),150-154.
MLA 何文轩,et al."基于YOLOv4-tiny的铁矿石品位识别技术研究".金属矿山 .10(2021):150-154.
Files in This Item:
File Name/Size DocType Version Access License
基于YOLOv4-tiny的铁矿石品位识(1970KB)期刊论文出版稿开放获取CC BY-NC-SAView Application Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[何文轩]'s Articles
[荆洪迪]'s Articles
[柳小波]'s Articles
Baidu academic
Similar articles in Baidu academic
[何文轩]'s Articles
[荆洪迪]'s Articles
[柳小波]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[何文轩]'s Articles
[荆洪迪]'s Articles
[柳小波]'s Articles
Terms of Use
No data!
Social Bookmark/Share
File name: 基于YOLOv4-tiny的铁矿石品位识别技术研究.pdf
Format: Adobe PDF
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.