zemli / Newly-Trained-Active-Shape-Models-with-Stasm

new Active Shape Model and its applications
2 stars 3 forks source link

It's very slow to run #1

Open GGBONEA opened 3 years ago

GGBONEA commented 3 years ago

I don't know why the Eyes_and_Head_Pose_Detection very slow to run to us source code ,but the app has run quickly more.

zemli commented 3 years ago

Hi, could you pls describe more about how you run the source code? It would be helpful to find the reason.

GGBONEA commented 3 years ago

Thank you for your reply,my test environment is VS2019+opencv249 。I use your 194ASM model to check eyes just 4 fps。but your 194app have 20 or 30 fps. i just use Eyes_and_Head_Pose_Detection. PS. could you tell me how to train an asm model

zemli commented 3 years ago

As the IDE occupies a lot of memory resources, it would be slow to run it with your IDE. And different VS settings influent debugging process differently.

GGBONEA commented 3 years ago

Thanks,i will try to change my IDE settng. could you teach me how to get a ASM model from Helen194 dataset?

zemli commented 3 years ago

Please refer to The MUCT landmarked face database. Pattern Recognition Association of South Africa, 201(0). The author shared the source code for model training.