# Extra
## Ground-Truth:
- ### convert `xml` to our format:
1) Insert ground-truth xml files into **ground-truth/**
2) Run the python script: `python convert_gt_xml.py`
- ### convert YOLO to our format:
1) Add class list to the file `class_list.txt`
2) Insert ground-truth files into **ground-truth/**
3) Insert images into **images/**
4) Run the python script: `python convert_gt_yolo.py`
- ### convert keras-yolo3 to our format:
1) Add or update the class list to the file `class_list.txt`
2) Use the parameter `--gt` to set the **ground-truth** source.
3) Run the python script: `python3 convert_keras-yolo3.py --gt <gt_file_path>`
1) Supports only python 3.
2) This code can handle recursive annotation structure. Just use the `-r` parameter.
3) The converted annotation is placed by default in a new from_kerasyolo3 folder. You can change that with the parameter `-o`.
4) The format is defined according with github.com/qqwweee/keras-yolo3
## Predicted:
- ### convert darkflow `json` to our format:
1) Insert result json files into **predicted/**
2) Run the python script: `python convert_pred_darkflow_json.py`
- ### convert YOLO to our format:
After runnuning darknet on a list of images, e.g.: `darknet.exe detector test data/voc.data yolo-voc.cfg yolo-voc.weights -dont_show -ext_output < data/test.txt > result.txt`
1) Copy the file `result.txt` to the folder `extra/`
2) Run the python script: `python convert_pred_yolo.py`
- ### convert keras-yolo3 to our format:
1) Add or update the class list to the file `class_list.txt`
2) Use the parameter `--predicted` to set the **prediction** source.
3) Run the python script: `python3 convert_keras-yolo3.py --pred <prediction_file_path>`
1) Supports only python 3.
2) This code can handle recursive annotation structure. Just use the `-r` parameter.
3) The converted annotation is placed by default in a new from_kerasyolo3 folder. You can change that with the parameter `-o`.
4) The format is defined according with github.com/gustavovaliati/keras-yolo3
## Remove specific char delimiter from files
E.g. remove `;` from:
`<class_name>;<left>;<top>;<right>;<bottom>`
to:
`<class_name> <left> <top> <right> <bottom>`
In the case you have the `--ground-truth` or `--predicted` files in the right format but with a specific char being used as a delimiter (e.g. `";"`), you can remove it by running:
`python remove_delimiter_char.py --char ";" --ground-truth`
## Find the files that contain a specific class of objects
1) Run the `find_class.py` script and specify the **class** as argument, e.g.
`python find_class.py chair`
## Remove all the instances of a specific class of objects
1) Run the `remove_class.py` script and specify the **class** as argument, e.g.
`python remove_class.py chair`
## Rename a specific class of objects
1) Run the `rename_class.py` script and specify the `--current-class-name` and `--new-class-name` as arguments, e.g.
`python rename_class.py --current-class-name Picture Frame --new-class-name PictureFrame`
## Rename all classes by replacing spaces with delimiters
Use this option instead of the above option when you have a lot of classes with spaces.
It's useful when renaming classes with spaces become tedious (because you have a lot of them).
1) Add class list to the file `class_list.txt` (the script will search this file for class names with spaces)
2) Run the `remove_space.py` script and specify the `--delimiter` (default: "-") and `--yes` if you want to force confirmation on all yes/no queries, e.g.
`python remove_space.py --delimiter "-" --yes`
## Intersect ground-truth and predicted files
This script ensures same number of files in ground-truth and predicted folder.
When you encounter file not found error, it's usually because you have
mismatched numbers of ground-truth and predicted files.
You can use this script to move ground-truth and predicted files that are
not in the intersection into a backup folder (backup_no_matches_found).
This will retain only files that have the same name in both folders.
1) Prepare `.txt` files in your `ground-truth` and `predicted` folders.
2) Run the `intersect-gt-and-pred.py` script to move non-intersected files into a backup folder (default: `backup_no_matches_found`).
`python intersect-gt-and-pred.py`
没有合适的资源?快使用搜索试试~ 我知道了~
snowman_yolo.rar
共596个文件
jpg:540个
py:30个
txt:9个
3星 · 超过75%的资源 需积分: 12 21 下载量 160 浏览量
2019-10-31
21:01:52
上传
评论
收藏 142.16MB RAR 举报
温馨提示
包含五百张训练样本和标签 tensorflow实现的yolov3 可搭配博客一起食用:https://blog.csdn.net/zmdsjtu/article/details/102845590
资源推荐
资源详情
资源评论
收起资源包目录
snowman_yolo.rar (596个子文件)
.gitignore 350B
events.out.tfevents.1572520077.HIH-D-14396 20.45MB
Box-Clustering.ipynb 1.02MB
road.jpeg 150KB
16693bf3d1d57fd3.jpg 981KB
3683237c9b797cd9.jpg 901KB
f3e77bd994850cdd.jpg 858KB
a4aaf955c19249e3.jpg 852KB
d9342ef381b5566a.jpg 826KB
339d38dddbd9f2f8.jpg 799KB
4cc9cf297d451be8.jpg 775KB
3430a7f24769b915.jpg 756KB
f5ae7645ef477d63.jpg 741KB
b6e16317ebbc944a.jpg 723KB
d46d4babf3285bf0.jpg 717KB
5ee97de6ada6efb9.jpg 712KB
b355a4e60eaed7ee.jpg 707KB
78e49257aec66314.jpg 706KB
0e41dc6b9e19444e.jpg 698KB
3850808ed2ab39f6.jpg 679KB
22c8d342fb892723.jpg 675KB
9d6d40f9ded757c7.jpg 667KB
a8249aade58b45c3.jpg 653KB
3247b1a2a22984fa.jpg 652KB
e93ab9b6411919e8.jpg 651KB
199cb281f2729b36.jpg 647KB
096914d22b2fbefa.jpg 644KB
42c7abbf2f228818.jpg 638KB
225dde21578d24ab.jpg 638KB
4dc34f85cfcbe3fd.jpg 631KB
0e15dd0c923c4626.jpg 630KB
0edc153d56a94bc7.jpg 623KB
a8f73b8ba76d32b5.jpg 608KB
01e7925576b00f6e.jpg 607KB
131bfc4997b4cfc7.jpg 595KB
2e846f89b5cf2559.jpg 587KB
20c82cfbf1247b58.jpg 584KB
7756d8974baff10a.jpg 582KB
58d25c4356a8fe37.jpg 578KB
3428b48dbd61dbd3.jpg 574KB
02758ed6b07b53cb.jpg 573KB
1e006d2a62a60a1a.jpg 560KB
34bd2300214b02ea.jpg 559KB
4733f751a1c0256c.jpg 558KB
e749043148f8681a.jpg 553KB
e12a48158009f830.jpg 547KB
9f5dbb32a037b71c.jpg 540KB
5d8f0e6f23cb2e67.jpg 531KB
b7897d9cc26388a5.jpg 523KB
24536c0270b45a5e.jpg 499KB
05abe73bee5e7b05.jpg 498KB
4ec038549e6c8329.jpg 495KB
d5ad648a554526fb.jpg 491KB
c4de7fcfd596ad5a.jpg 489KB
228dd5befd0f6006.jpg 489KB
2346479af2e864fa.jpg 488KB
219fcd86e9ca5d1a.jpg 486KB
022583764052e685.jpg 483KB
059d36652a6d1f05.jpg 483KB
59c54e75b35156dc.jpg 478KB
4f3631ab5654e326.jpg 478KB
2647da06c1f82937.jpg 472KB
3b3c0082a641b7d2.jpg 471KB
0b8d6c8458eb9cab.jpg 470KB
1c1a0f078db41d83.jpg 468KB
74f54e02bc704733.jpg 462KB
4477d8f41a075102.jpg 462KB
1a61e3e20053d0dc.jpg 462KB
b26f003a4e26d1be.jpg 459KB
8c460287e2659772.jpg 456KB
8a2cd00033785bd9.jpg 456KB
022cdc857c47d623.jpg 456KB
f527f7ddfe7eb806.jpg 455KB
0011879633d8ef0c.jpg 455KB
28233eb3f3d68401.jpg 455KB
02a1bb6036a6a99c.jpg 452KB
86a27d6b8800cf93.jpg 452KB
9757662c58237f8c.jpg 451KB
26bb9909a07ccc67.jpg 451KB
1f102c1e237dba6a.jpg 450KB
189daa915d62ca11.jpg 445KB
325d1af1e1b42ee2.jpg 443KB
4d1c088e919e1b4c.jpg 441KB
24abc9e2759b861a.jpg 439KB
1e02d8dea8217323.jpg 438KB
2a13c384d93e782d.jpg 433KB
3b064f87aff0c1ce.jpg 433KB
ec4329992a99afe1.jpg 431KB
17521523860e1039.jpg 428KB
4c0d66d162726eb5.jpg 426KB
b65e325ff4af3937.jpg 425KB
c5dc95ef9f40d8d0.jpg 425KB
ce5243c7cec58041.jpg 425KB
0cdd83dd219791e0.jpg 419KB
28b7a48928d1b80e.jpg 418KB
c413b7242a87e133.jpg 412KB
0b65ff98d29f6d77.jpg 412KB
3a53fcf2aeaaeae3.jpg 410KB
313385cd5a2aadd8.jpg 410KB
0f3bffe8b2dbb660.jpg 409KB
共 596 条
- 1
- 2
- 3
- 4
- 5
- 6
资源评论
- lsldd2020-02-23感谢楼主分享
- cmengwei2021-05-10分数太多了
朱铭德
- 粉丝: 1w+
- 资源: 39
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功