Skip to content

Spphire/RM-labeling-tool

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

click "Play now!" link above, you can quickly get vast datasets of Energy Mechanism and Armor

一些有趣的小工具./Tool/pycontroller,批量生产数据集的过程中随机替换你提供的背景图片来达到数据增强的效果,详见wiki

⚠️ 通知 2023/3/29 **官方更改未激活扇叶灯效,现已修改

新增功能 TAB键打开关闭功能面板

⚠️ 通知 2023/3/28 1:00前获取的数据集xywh有问题,但关键点是正确的,现已修复

⚠️ 之前获取的数据集可用Tool下脚本进行修复

操作指南

如果你想把获得的标注文件转换为yolo-face的格式,你可以参考转换工具

输出已为yoloface格式

TAB键打开关闭功能面板

P键获取截图,强烈建议 在浏览器设置中关闭(下载前询问),并设置好合理的下载路径

你会获得四个文件,一张原图,一张画框的图,两个标注文件(其中一个是txt,一个是json,两种格式,json可读性高一些,根据需要选一种用)

按住左键 才能用WASD和鼠标控制移动

左键画面即可隐藏鼠标光标,esc或空格退出。

R开关能量机关旋转,T随机让能量机关亮灯(随机状态)

F可以在当前位置生成一个装甲板

屏幕中心瞄准装甲板时会使其处于高亮状态(不会影响保存的截图)

高亮状态下,按住右键可以让装甲板跟随自己移动

高亮状态下, 0234568键可切换装甲板贴纸

高亮状态下, C键切换红蓝灯

💞 看到这了给颗星吧

标签编号解释

对于能量机关来说,最终label=color*3+tag

color 0: 蓝, 1 红

tag 0:R,1:未激活,2:已激活

对于装甲板来说,最终label=color*9+tag

color 0: 蓝, 1:红, 2:灰, 3:紫

tag 0:哨兵,1:英雄,2:工程,3:步兵3,4:步兵4,5:步兵5,6:前哨站,7、8:大小基地装甲板

Instrcutions

You can refer to convert tool if you want to convert dataset from my project to Yolo-face format.

already yoloface format

Moving:

·Hold the left button of mouse and

·Press the WASD keys and cooperate with the mouse

Controling:

For Energy Mechanism

·Press the R keys to turn on or off the rotation of Energy Mechanism

·Press the T keys to randomly set the state of the fan blades of Energy Mechanism (OFF,UNACTIVATED,ACTIVATED)

For Armor Plate

·Press the F keys to create a new armor plate on your current position

·Move the center of the screen to the armor plate, and that the armor can be controled(with green outline)

Press the 0,2,3,4,5,6,8 keys to set the tag of the armor controled(with green outline)

Press the C keys to switch the light color of the armor controled(with green outline) red or blue

Hold on pressing right mouse button, you can grab the armor with you

Photoing:

·Press the P keys to get a screenshot and it will be labeled automatically

(Tips: Turn off "Asking before downloading" in your browser settings and set the default download path so that You don't need to manually confirm the download every time you photo)

You will get 4 files, one is .jpg which is raw picture, two are .json and .txt which includes label information (2 format), one is .png which visualizes the label information to check if there's something wrong.

explanation of label information:

For Energy Mechanism:

0: blue R, 1: blue Unactivated, 2: blue Activated, 3: red R, 4: red Unactivated, 5: red Activated .

For Armor Plate:

final label = color*9+tag

with color, 0: blue, 1: red, 2: grey, 3: purple

with tag, 0: sentry, 1: hero, 2: engineer, 3: infantry3, 4: infantry4, 5: infantry5, 6: outpost, 7: base(small), 8: base(big)

labeling performance

draw2023_3_29_15_56_28_730

draw2023_3_29_15_56_46_501

outdated:

draw2023_3_16_17_0_42_393

draw2023_3_8_17_40_33_832

About

It's a simulator based on Unity for RoboMaster. You can use it to get some labeled dataset for deep learning

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published