clmtrackr
clmtrackr copied to clipboard
improving performance on mobile
Hi @auduno
I've been working with your library for facing detection supposed to work on mobile. Is there is any way to improve performance on mobile devices ? I've already try to reduce the searchWindow value, activating WebGL, I'm also calling the track function on a video source at a reduced frame rate, which already improve the performance a bit.
If you can find to any other way that would be great.
Thanks.
Hi @auduno and @guilanier
Did you get any results related to this issue? I've tried the same things @auduno did, but the performance is still bad in mobile devices.
Thanks.
So anything happened on performance or on what to do to improve them on mobile ? I have 160 fps on a laptop, but 7fps on mobile (and I think it's just 0.2 fps in real)
The biggest problem I'm facing with android is uploading the image to begin with. On Samsung phones the image displays in landscape even though it was takin in portrait front facing camera. So all images are rotated the wrong way. Anyone find away around this ??
I use this library to auto-rotate uploaded images based on exif data. Works well on iOS. I assume it works on Android as well.
https://github.com/blueimp/JavaScript-Load-Image
Code for using web workers during initial face detection has now been added, this might help with some of the latency issues.
Still getting ~5fps on Google Pixel - with or without web workers - on an uploaded image. Are there any under-the-hood tweaks that can be done to improve this? I don't mind getting my hands dirty but I can't figure what's causing this! Seems to be less of an issue on iOS for some reason.