how to prepare training data for V-J face detector?

[Watebear]

Hello, I am learning this algothrim. Your codes help me a lot.
But I don't know how tor prepare the training data for this project.
Thank you very much, sincerely.

[johan456789]

.
├── tests
├── trainingdata
│   ├── faces
│   │   └── test
│   └── nonfaces
│       └── test
└── violajones
    └── __pycache__

I'm not the author, but you should put training data in trainingdata/faces and trainingdata/nonfaces, and put testing data in trainingdata/faces/test and trainingdata/nonfaces/test.

[jakewilliami]

Alternatively to @johan456789's suggestion, edit these lines.

[musgan]

Alternatively to @johan456789's suggestion, edit these lines.

pos_training_path = 'trainingdata/faces'
neg_training_path = 'trainingdata/nonfaces'
pos_testing_path = 'trainingdata/faces/test'
neg_testing_path = 'trainingdata/nonfaces/test'

where can i get training data ? Can you share data exists for pre-training and make it simple. I just want use for detect face, just it.

thankyou

[jakewilliami]

@muhusgan The dataset used for testing by @Simon-Hohberg can be found here (source). There are many other datasets and resources available for face detection, but the one linked is nice because there are lots of very small images.

If you are on a Unix machine (macOS or Linux), or have access to the command line, you can simply run

cd /tmp/
git clone https://github.com/INVASIS/Viola-Jones.git/
cd - # or change working directory to wherever you need to be
mv /tmp/Viola-Jones/data/trainset/ ./
mv /tmp/Viola-Jones/data/testset/ ./

Now your data files should look like this:

pos_training_path = 'trainset/faces'
neg_training_path = 'trainset/nonfaces'
pos_testing_path = 'testset/faces'
neg_testing_path = 'testset/nonfaces'

It doesn't really matter how your data is structured, as long as the algorithm knows where to find it.

Let me know if you need further help with this.