Test data for the DNN Face module #903
Conversation
…es under cv/cascadeandhog/images
|
What is the source of used images? license? |
The used images are selected from Labeled Faces in the Wild database. Labeled Faces in the Wild is a public benchmark for face verification, also known as pair matching. And it was published to help the research community make advances in face verification. We can not find any license of this database from its homepage. For more details about this database, please see here. |
testdata/dnn/download_models.py
Outdated
| name='YuNet', | ||
| url='https://github.com/ShiqiYu/libfacedetection.train/raw/master/tasks/task1/onnx/yunet.onnx', | ||
| sha='49c52f484b1895e8298dc59e37f262ba7841a601', | ||
| filename='onnx/models/yunet.onnx'), |
There was a problem hiding this comment.
/master/
Use exact commit hash instead of master.
onnx/models/yunet.onnx
If you plan to update model in the future (which is happened recently) please add something like -YYYYMM suffix to stored file name (e.g, onnx/models/yunet-202109.onnx).
Target location for download_models.py files is assumed as a shared storage without any support for versioning or file's updates. It is shared between different branches or OpenCV releases so its content should be persistent (you can add, but can't remove/replace).
There was a problem hiding this comment.
Sorry for the late response. These have been done in the latest commit.
Related PR: opencv/opencv#20422