Yolo_Label/README.md

75 lines
3.0 KiB
Markdown
Raw Normal View History

2018-10-29 11:52:37 +03:00
# Yolo_label
2018-10-29 07:49:17 +03:00
## WHAT IS THIS?!
Reinventing The Wheel!!!!
2018-10-29 08:38:26 +03:00
![1_hfyjxxcfingbcyzcgksaiq](https://user-images.githubusercontent.com/35001605/47629997-b47aa200-db81-11e8-8873-71ae653563e0.png)
2018-10-29 09:09:07 +03:00
In the world, there are many good image labeling tools for object detection. -e.g. , ([Yolo_mark](https://github.com/AlexeyAB/Yolo_mark), [BBox-Label-Tool](https://github.com/puzzledqs/BBox-Label-Tool), [labelImg](https://github.com/tzutalin/labelImg)).
2018-10-29 07:49:17 +03:00
2018-10-29 09:09:07 +03:00
But... I reinvented it...
2018-10-29 08:38:26 +03:00
2018-10-29 10:36:50 +03:00
## WHY DID YOU REINVENT THE WHEEL? ARE YOU STUPID?
2018-10-29 08:38:26 +03:00
2018-10-29 09:09:07 +03:00
When I use the pre-existing program to annotate a training set for YOLO V3, I am sooooooooooo bored...
2018-10-29 08:38:26 +03:00
2018-10-29 10:15:13 +03:00
So I thought about why it is so fxxking bored??
2018-10-29 09:09:07 +03:00
And I found that anwer.
2018-10-29 07:49:17 +03:00
2018-10-29 10:15:13 +03:00
The answer is that pre-existing programs are not **sensitive**.
2018-10-29 07:49:17 +03:00
2018-10-29 12:21:42 +03:00
So I decided to make a **sensitive** image labeling tool for object detection.
2018-10-29 10:15:13 +03:00
## SHOW ME YOUR SENSITIVE IMAGE LABELING TOOL!!
2018-10-29 19:02:41 +03:00
It's the **SENSITIVE** image labeling tool for object detection!
2018-10-29 11:48:20 +03:00
2018-10-29 19:02:41 +03:00
![image](https://user-images.githubusercontent.com/35001605/47644151-bd369c80-dbb0-11e8-9505-d7ada78911c2.png)
2018-10-29 08:38:26 +03:00
2018-10-30 06:00:08 +03:00
![ezgif-5-85c0613befb3](https://user-images.githubusercontent.com/35001605/47693025-3d094900-dc3b-11e8-9d96-ea2f7e37cf7a.gif)
![ezgif-5-c57fe634145b](https://user-images.githubusercontent.com/35001605/47693028-3f6ba300-dc3b-11e8-9e22-70c448d59398.gif)
2018-10-29 11:44:52 +03:00
2018-10-29 10:31:00 +03:00
## HMM... I SAW THIS DESIGN SOMEWHERE
2018-10-29 11:46:42 +03:00
I referred to [the website of Joseph Redmon](https://pjreddie.com/darknet/
2018-10-29 11:44:52 +03:00
) who invented the [YOLO](https://www.youtube.com/watch?v=MPU2HistivI).
2018-10-29 12:21:42 +03:00
2018-10-29 10:36:08 +03:00
![redmon2](https://user-images.githubusercontent.com/35001605/47635529-a1270100-db98-11e8-8c03-1dcea7c77d1d.PNG)
2018-10-29 12:30:26 +03:00
# TUTORIAL / USAGE
2018-10-29 13:23:44 +03:00
1. Put your .jpg, .png -images into some directory
(In this tutorial I will use the Kangarooo and the Raccoon Images. These images are in the 'Samples' folder.)
2018-10-29 12:21:42 +03:00
![image](https://user-images.githubusercontent.com/35001605/47640641-5791e280-dba7-11e8-8709-83fba08bbbcd.png)
2018-10-30 06:03:26 +03:00
2. Put names of objects, one for each line in some file( .txt, .names) and save
2018-10-29 12:21:42 +03:00
2018-10-29 13:23:44 +03:00
![image](https://user-images.githubusercontent.com/35001605/47643863-0803e480-dbb0-11e8-87b1-2becf72f4a43.png)
2018-10-29 10:15:13 +03:00
2018-10-29 19:02:41 +03:00
3. Run Yolo label!
2018-10-29 08:38:26 +03:00
2018-10-29 19:02:41 +03:00
![image](https://user-images.githubusercontent.com/35001605/47644151-bd369c80-dbb0-11e8-9505-d7ada78911c2.png)
2018-10-30 06:03:26 +03:00
4. Click the button 'Open Files'and open the directory that you saved your custom images and object name lists file that you saved at step 2.
2018-10-30 06:37:00 +03:00
![image](https://user-images.githubusercontent.com/35001605/47694149-d20e4100-dc3f-11e8-9d97-fba87a6e1b5a.png)
5. And... Drag and Drop... Welcome to Hell... I really hate this work in the world
2018-10-30 10:25:24 +03:00
![ezgif-5-805073516651](https://user-images.githubusercontent.com/35001605/47698872-5bc80980-dc54-11e8-8984-e3e1230eccaf.gif)
6. End
## Shorcuts
| Key | Action |
|---|:---:|
| `A` | Save and Next Image |
| `D, Space` | Save and Next Image |
| `W` | 위치 상 부모(조상)요소를 기준으로 배치 |
| `S` | 브라우저 창을 기준으로 배치 |
| `Ctrl + S` | 브라우저 창을 기준으로 배치 |
| `Ctrl + C` | 브라우저 창을 기준으로 배치 |
## Mouse Events