That's cool, I have also tried the same with cv2, but we can't expect 100% of face match, with a lower quality camera.
But I have tried in a different way, like setting images in a particular folder with person name, so that it trains the pictures in that folder with folder name, draws a graph on face so while reading the video and validates the face from example% to example%,
Appreciate that but actually in some ways you've done better like directly reading from the video and we'll look forward to change according to your suggestions. And yeah the camera quality is lower so accuracy is not as expected.
but my project needed a good ammount of pictures, atleast 4-5 pics to get a decent recognition, pictures to train, so it wont get confused and give other details. and you have to use PiL for image processing functionality
That's cool, I have also tried the same with cv2, but we can't expect 100% of face match, with a lower quality camera.
But I have tried in a different way, like setting images in a particular folder with person name, so that it trains the pictures in that folder with folder name, draws a graph on face so while reading the video and validates the face from example% to example%,
my-code
And finally using mask for face like setting hue, saturation, so and so for making it easier to read the video.
My project might not be as good as you guys but I just wanna know, this is a proper or have to do change
Appreciate that but actually in some ways you've done better like directly reading from the video and we'll look forward to change according to your suggestions. And yeah the camera quality is lower so accuracy is not as expected.
but my project needed a good ammount of pictures, atleast 4-5 pics to get a decent recognition, pictures to train, so it wont get confused and give other details. and you have to use PiL for image processing functionality
Yes, that's true. I did look into that in your repo.