Skip to content

Use yolov3 to detect the position of the object, and measure the size of the object with the point cloud data collected by zivid.

Notifications You must be signed in to change notification settings

BIG-DD/Measurement_base_on_yolov3

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

项目说明

本项目使用yolov3对目标在2D图像中的位置进行检测,获取检测框的4个顶点坐标。然后将坐标映射到2D图像对应的3D数据中,根据3D数据计算出物体的真实尺寸。

环境准备

win10.

torch 1.6.0.

python3.6

训练

数据集准备

自己的数据集标记好后, png文件放入 datasets/DOTA_data/images中,xml文件放入annotation中。 然后运行 data_generate_txtfile.pyImageSet中生成train.txt val.txt test.txt三个文件,再运行xml_txt.pyannotations 中的xml文件转换成txt文件存在labelTxt中。

config

cfg/yolov3.cfg的603、610、689、696、776、783行进行修改,1修改filters的数量,2修改classes的数量(即目标检测的类别数).

cfg/icdar.names进行修改,写入标签的名称,名称顺序与datasets/DOTA_data/xml_txt.py中classes的名称顺序一致。

训练网络

运行train.py训练网络

预测模型

运行predict.py,修改自己训练好的网络权重位置和预测结果存放位置。

预测和测量

运行detect.py对物体进行尺寸测量,测量时依赖zivid 相机采到的数据。

About

Use yolov3 to detect the position of the object, and measure the size of the object with the point cloud data collected by zivid.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages