model with landmarks

[natxopedreira]

Hello

Thanks for this repo, is very fast!!

Im using the models you provide with opencv and openvino and works great!!!

But seems to not have landmarks..... how can i get landmarks?? I tried converting v2 caffe model but again no landmarks on output layer

Also your photo results seems much better than the results im getting? Maybe the model is not update?

Regards

[ShiqiYu]

Only the newest v3 model has 5 landmarks

On Wed, May 20, 2020 at 9:23 PM natxopedreira notifications@github.com wrote:

Closed #252 https://github.com/ShiqiYu/libfacedetection/issues/252.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/ShiqiYu/libfacedetection/issues/252#event-3356444595, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABWR4HITPUR7GUCOZMZKTIDRSPKVDANCNFSM4NDSCHVQ .

--

Prof. Shiqi YU (于仕琪)

Department of Computer Science and Engineering,

Southern University of Science and Technology,

Shenzhen, China.

[natxopedreira]

Thank you, and how can i get that model?

[ShiqiYu]

https://github.com/ShiqiYu/libfacedetection.train/tree/master/tasks/task1

[natxopedreira]

Thanks again, the output of that model seems more complex than your openvino version and code is too much complex for me to understand how is done the conversion inside your library.

Is there any place where some kind of documentation on what the model output is exist?

[ShiqiYu]

I also forgot how to run it using OpenVINO. You can find object examples in OpenVINO. They should be similar.

On Thu, May 21, 2020 at 4:42 AM natxopedreira notifications@github.com wrote:

Thanks again, the output of that model seems more complex than your openvino version and code is too much complex for me to understand how is done the conversion inside your library.

Is there any place where some kind of documentation on what the model output is exist?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/ShiqiYu/libfacedetection/issues/252#issuecomment-631714910, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABWR4HNMCCGH6T7B5OFHELTRSQ6BTANCNFSM4NDSCHVQ .

[natxopedreira]

Thank you so much for your response and for help!!

I convert the model to openvino and the output is very extrange.

Created a gist with the output, i will be very glad if you can take a look https://gist.github.com/natxopedreira/111d51252b1e759c67738869885d4bc3

I converted /opt/intel/openvino_2020.2.117/deployment_tools/model_optimizer/mo.py --input_model libfacedetection.train-master/tasks/task1/onnx/facedetectcnn_320.onnx

I should add that with that python code, models usually output in a comprensible shape