Skip to content

Commit

Permalink
Update README.md
Browse files Browse the repository at this point in the history
  • Loading branch information
heinrichI authored Aug 16, 2019
1 parent 77efc44 commit 37c2685
Showing 1 changed file with 20 additions and 16 deletions.
36 changes: 20 additions & 16 deletions README.md
Original file line number Diff line number Diff line change
@@ -1,28 +1,32 @@
# FaceRecognitionWPF
FaceRecognitionWPF

A wrapper over FaceRecognitionDotNet to search for known people in directories.
![FaceRecognitionWPF Main window](/img/MainWindow.jpg?raw=true "Main window")

A wrapper over FaceRecognitionDotNet to search for famous people in directories.
A wrapper over FaceRecognitionDotNet to search for known people in directories.
Uses:
FaceRecognitionDotNet (Porting face_recognition (by Adam Geitgey) by C #)
DlibDotNet (.NET wrapper for DLib written in C #)
LiteDB - A lightweight embedded .NET NoSQL document store in a single datafile
* FaceRecognitionDotNet (Porting face_recognition (by Adam Geitgey) by C #)
* DlibDotNet (.NET wrapper for DLib written in C #)
* LiteDB - A lightweight embedded .NET NoSQL document store in a single datafile

Usage:
1. Create directories with the names of people in the training directory and throw photos with faces in them.
![TrainDirectory.jpg](/img/TrainDirectory.jpg?raw=true)
2. Indicate the folder with the trained model for describing faces. You can download it from https: //github.com/ageitgey/face_recognition_models
3. Select a folder to search and proximity level.
4. The form displays faces with a level of proximity less than a given. You can double-click on this image or copy it to a training folder to improve your search.
You can also use the Show another class and distance more than button to show, on the contrary, face images for which the threshold is greater than the threshold for the specified class. In order to find people who are not perceived as belonging to this person.
2. Indicate the folder with the trained model for describing faces. You can download it from https://github.com/ageitgey/face_recognition_models
3. Select a directory to search and distance treshold (for KNN search).
4. The form displays faces with a distance less than a given. You can double-click on this image or copy it to a training folder to improve your search.
You can also use the "Show another class and distance more than" button to show, on the contrary, face images for which the threshold is greater than the threshold for the specified class. In order to find people who are not perceived as belonging to this person.

Оболочка над FaceRecognitionDotNet, для поиска известных людей в директориях.
Использует:
FaceRecognitionDotNet (Porting face_recognition (by Adam Geitgey) by C#)
DlibDotNet (.NET wrapper for DLib written in C#)
LiteDB - A lightweight embedded .NET NoSQL document store in a single datafile
1. Создаете в директории тренировки директории с именами людей и кидаете в них фотографии с лицами.
2. Указываете папку с тренированной моделью описания лиц. Скачать можно сhttps://github.com/ageitgey/face_recognition_models
3. Выбираете папку для поиска и уровень близости.
* FaceRecognitionDotNet (Porting face_recognition (by Adam Geitgey) by C#)
* DlibDotNet (.NET wrapper for DLib written in C#)
* LiteDB - A lightweight embedded .NET NoSQL document store in a single datafile

Использование:
1. Создайте в директории тренировки директории с именами людей и кидаете в них фотографии с лицами.
![TrainDirectory.jpg](/img/TrainDirectory.jpg?raw=true)
2. Указываете папку с тренированной моделью описания лиц. Скачать можно с https://github.com/ageitgey/face_recognition_models
3. Выбираете папку для поиска и уровень близости (для KNN поиска).
4. В форме отображаются лица, с уровнем близости меньше заданного. Можно открыть это изображени по двойному клику или скопировать в папку для тренировки для улучшения поиска.
Еще можно по кнопке Show another class and distance more than показать наоборот изображения лица для которых порог больше заданного для указанного класса. Для того чтобы найти лица которые не воспринимаеются как принадлежащие этому человеку.
Еще можно по кнопке "Show another class and distance more than" показать наоборот изображения лица для которых порог больше заданного для указанного класса. Для того чтобы найти лица которые не воспринимаеются как принадлежащие этому человеку.

0 comments on commit 37c2685

Please sign in to comment.