Model fitting using keypoint regression
Keypoints are predicted in an image. Predictions are generated for each of the keypoints of an image as a 2D random variable, normally distributed with location (x, y) and standard deviation sigma. A neural network is trained to maximize a log-likelihood that samples from each of the predicted keypoints equal a ground truth. The trained neural network is used to predict keypoints of an image without generating a heatmap.
1. A method for predicting keypoints by a computing system, the method comprising:
receiving, by the computing system, data indicative of a plurality of images;
generating, by the computing system, predictions for keypoints of the plurality of images as 2D random variables, normally distributed with location (x, y) and standard deviation sigma;
training, by the computing system, a neural network to maximize a log-likelihood that samples from each of the predicted keypoints equal a ground truth by minimizing a sum of Gaussian negative log likelihoods; wherein the training comprises introducing a conjugate prior of a Gaussian distribution of uncertainty values for the predicted keypoints;
using the trained neural network to predict keypoints of a 3D image without generating a heatmap; and
based on the predicted keypoints of the 3D image, outputting a fitted 3D model for rendering on a display device.
2. The method of claim 1 , further comprising fitting a parametric model to the predicted keypoints.
3. The method of claim 2 , wherein an energy function for fitting the parametric model comprises
E
(
Φ
;
L
)
=
E
landmarks
︸
Data
term
+
E
identity
+
E
expression
+
E
joints
+
E
temporal
+
E
intersect
︸
Regularizers
wherein E(Φ; L) comprises the energy function, E landmarks is a data term and remaining terms use prior knowledge to regularize the fit of the parametric model.
4. The method of claim 1 , wherein distribution of uncertainty values is influenced during the training by introducing a suitable prior.
5. The method of claim 1 , wherein the image is of an object for which an a-priori model is available, the method further comprising preprocessing the image by:
running a sliding window over the image;
measuring average keypoint confidence for each window of the sliding window;
if a window with a high average keypoint certainty is not found, determining that the object is not in the image; and
otherwise, taking the window which reported a highest average keypoint confidence to contain the object.
6. The method of claim 1 , wherein the image is input from one or more of regular color (RGB) cameras, depth cameras, IR sensors, head-mounted cameras, event cameras, or web cameras.
7. The method of claim 1 , wherein the method is performed in combination with a model fitter that predicts intrinsic camera parameters.
8. The method of claim 7 , wherein the model fitter is configured for a single view where uncertainty of each landmark is taken into account.
9. The method of claim 7 , wherein the model fitter is configured for multiple views.
10. A computing system for fitting a model using observation data, the computing system comprising:
one or more processors; and
a computer-readable storage medium having computer-executable instructions stored thereupon which, when executed by the processor, cause the computing system to perform operations comprising:
receiving, by the computing system, data indicative of a plurality of images;
generating predictions for keypoints of the plurality of images as 2D random variables, normally distributed with location (x, y) and standard deviation sigma;
training a neural network to maximize a log-likelihood that samples from each predicted keypoint equal a ground truth by minimizing a sum of Gaussian negative log likelihoods; wherein the training comprises introducing a conjugate prior of a Gaussian distribution of uncertainty values for the predicted keypoints;
using the trained neural network to predict keypoints of a 3D image;
fitting a parametric model to the predicted keypoints; and
based on the predicted keypoints of the 3D image, outputting a fitted 3D model for rendering on a display device.
11. The computing system of claim 10 , wherein distribution of uncertainty values is influenced during the training by introducing a suitable prior.
12. The computing system of claim 10 , wherein the image is of an object, further comprising computer-executable instructions stored thereupon which, when executed by the processor, cause the computing system to perform operations for preprocessing the image, comprising:
running a sliding window over the image;
measuring average keypoint confidence for each window;
if a window with a high average keypoint certainty is not found, determining that the object is not in the image; and
otherwise, taking the window which reported the highest average keypoint confidence to contain the object.
13. The computing system of claim 10 , wherein the predicting and training is performed in combination with a model fitter that predicts intrinsic camera parameters.
14. The computing system of claim 13 , wherein the model fitter is configured for a single view where uncertainty of each landmark is taken into account.
15. The computing system of claim 10 , further comprising computer-executable instructions stored thereupon which, when executed by the processor, cause the computing system to perform operations comprising performing 3D reconstruction from multiple views from multiple cameras, where an uncertainty in each view is taken into account, and wherein extrinsic parameters of each camera are concurrently optimized.
16. The computer system of claim 15 , wherein the cameras are HMD cameras.
17. The computing system of claim 10 , computer-executable instructions stored thereupon which, when executed by the processor, cause the computing system to perform operations comprising using uncertainties of the neural network to estimate which parts of an object being tracked are visible.
18. A non-transitory computer-readable storage medium having computer-executable instructions stored thereupon which, when executed by a processor of a computing device, cause the computing device to perform operations comprising:
receiving data indicative of a plurality of images;
generating predictions for keypoints of the plurality of images as 2D random variables, normally distributed with location (x, y) and standard deviation sigma;
training a neural network to maximize a log-likelihood that samples from each predicted keypoint equal a ground truth by minimizing a sum of Gaussian negative log likelihoods; wherein the training comprises introducing a conjugate prior of a Gaussian distribution of uncertainty values for the predicted keypoints;
using the trained neural network to predict keypoints of a 3D image;
fitting a parametric model to the predicted keypoints; and
based on the predicted keypoints of the 3D image, outputting a fitted 3D model for rendering on a display device.
19. The non-transitory computer-readable storage medium of claim 18 , wherein the image is of an object for which an a-priori model is available, further comprising preprocessing the image by:
running a sliding window over the image;
measuring average keypoint confidence for each window of the sliding window;
if a window with a high average keypoint certainty is not found, determining that the object is not in the image; and
otherwise, taking the window which reported a highest average keypoint confidence to contain the object.
20. The non-transitory computer-readable storage medium of claim 18 , wherein the predicting and training is performed in combination with a model fitter that predicts intrinsic camera parameters.