- Marcar como nuevo
- Favorito
- Suscribir
- Silenciar
- Suscribirse a un feed RSS
- Resaltar
- Imprimir
- Informe de contenido inapropiado
When using the Face Detection I've noticed when you flip your system upside down (e.g. on a Yoga when using it in "Tent" mode) the engine no longer can detect faces reliably. Same issue when using a Tablet in Portrait mode.
Is there a way to tell the SDK what the orientation of the camera is to "flip" the image correctly before it does face detection?
- Etiquetas:
- Intel® RealSense™ Technology
Enlace copiado
- Marcar como nuevo
- Favorito
- Suscribir
- Silenciar
- Suscribirse a un feed RSS
- Resaltar
- Imprimir
- Informe de contenido inapropiado
Why don't you turn upside down the frame before giving it to the sdk?..it may work :)
Mike
- Marcar como nuevo
- Favorito
- Suscribir
- Silenciar
- Suscribirse a un feed RSS
- Resaltar
- Imprimir
- Informe de contenido inapropiado
That's actually sort of what I'm asking, is how do I "Give" the SDK a frame. I fail to see anywhere you can intercept a frame from the camera and flip it. That all seems to happen under the hood.
For example on the Emotion Detection you call (psuedocode below)..
while(true)
{
auto status = PXCSSenseManager.AcquireFrame(true);
auto emotions = PXCSSenseManager.QueryEmotion();
}
- Marcar como nuevo
- Favorito
- Suscribir
- Silenciar
- Suscribirse a un feed RSS
- Resaltar
- Imprimir
- Informe de contenido inapropiado
mmm, yeah, had a quick look through the emotion sample and found this call:
const PXCCapture::Sample *sample = putil->QueryEmotionSample();
It seems that the SDK is requesting the frame and the emotion processing at the same time...
Also, I didn't find any reference to an image in pxcemotion.h.
It feels strange that the API does not provide a way of inputting an image to the emotion processing part... maybe I am not looking in the right places?
- Marcar como nuevo
- Favorito
- Suscribir
- Silenciar
- Suscribirse a un feed RSS
- Resaltar
- Imprimir
- Informe de contenido inapropiado
I suppose a long-shot might be that you could set your program up with the Object Tracking feature, and program into it images of an upside down person's facial features so that when the camera sees those features (e.g an upside down mouth), it would work out that it should carry out the same instruction that the facial recognition feature would when the face was right-way-up.
- Marcar como nuevo
- Favorito
- Suscribir
- Silenciar
- Suscribirse a un feed RSS
- Resaltar
- Imprimir
- Informe de contenido inapropiado
After setting up the pipeline you can do something like
PXCCapture::Device *device = pp->QueryCaptureManager()->QueryDevice(); device->SetMirrorMode(PXCCapture::Device::MirrorMode::MIRROR_MODE_HORIZONTAL);
Where pp is a PXCSenseManager object returned by PXCSenseManager::CreateInstance();
However, I can't figure out how to use the SDK to flip an image vertically. Anyone have any suggestions?

- Suscribirse a un feed RSS
- Marcar tema como nuevo
- Marcar tema como leído
- Flotar este Tema para el usuario actual
- Favorito
- Suscribir
- Página de impresión sencilla