Generating a three-dimensional topography of a training environment
A method for adaptive user training and gaming utilizing a media player is disclosed that includes acquiring images from cameras surrounding a training environment and attempting to segment a ball and a human in each of the images, wherein when failing to segment either the ball or the human in a first image, the first image is rejected from further processing. The operations additionally include creating masks of the ball and the human, extracting features for both the ball and the human in each of the remaining images, and computing surface information of the training environment using photoclinometry. For each of the remaining images, the operations includes computing a distance between the ball and the human. Additionally, the operations include identifying an image having the greatest distance between the ball and the human and computing a three-dimensional topography from the identified image.
1. A computerized method comprising:
acquiring data from a plurality of cameras surrounding a training environment, the data including a plurality of images;
attempting to segment a ball and a human in each of the plurality of images, wherein responsive to failing to segment either the ball or the human in a first image, rejecting the first image from further processing;
creating a mask of the ball and a mask of the human;
extracting features for both the ball and the human in each of the plurality of images that were not rejected from further processing;
computing surface information of the training environment using a photoclinometry process;
for each of the plurality of images that were not rejected from further processing, computing a distance between the ball and the human;
identifying an image of the plurality of images having a greatest distance between the ball and the human; and
computing a three-dimensional topography from the identified image.
2. The computerized method of claim 1 , wherein the distance between the ball and the human is a Euclidean distance.
3. The computerized method of claim 1 , wherein the photoclinometry process includes operations that transform a two-dimensional (2D) image into a surface map representing varying depth within the 2D image.
4. The computerized method of claim 1 , wherein the data acquired from the plurality of cameras includes images or videos that capture the ball, wherein a ball-throwing machine has imparted motion to the ball.
5. The computerized method of claim 1 , wherein computing the three-dimensional topography from the identified image includes utilization of a triangulation technique to determine a position of the ball and a position of the human in three-dimensional space.
6. The computerized method of claim 1 , wherein extracting the features for both the ball and the human in each of the plurality of images that were not rejected from further processing includes:
recreating a three-dimensional relief of the user and of the ball, and
extracting pixel-based features of the ball and of the human from the three-dimensional relief, wherein the pixel-based features extracted from the three-dimensional relief are used in determining the greatest distance between the ball and the human.
7. The computerized method of claim 1 , further comprising:
determining an image to be displayed to the human based on the three-dimensional topography computed from the identified image.
8. A computing device, comprising:
one or more processors; and
a non-transitory computer-readable medium having stored thereon instructions that, when executed by the processor, cause the processor to perform operations including:
acquiring data from a plurality of cameras surrounding a training environment, the data including a plurality of images,
attempting to segment a ball and a human in each of the plurality of images, wherein responsive to failing to segment either the ball or the human in a first image, rejecting the first image from further processing,
creating a mask of the ball and a mask of the human,
extracting features for both the ball and the human in each of the plurality of images that were not rejected from further processing,
computing surface information of the training environment using a photoclinometry process,
for each of the plurality of images that were not rejected from further processing, computing a distance between the ball and the human,
identifying an image of the plurality of images having a greatest distance between the ball and the human, and
computing a three-dimensional topography from the identified image.
9. The computing device of claim 8 , wherein the distance between the ball and the human is a Euclidean distance.
10. The computing device of claim 8 , wherein the photoclinometry process includes operations that transform a two-dimensional (2D) image into a surface map representing varying depth within the 2D image.
11. The computing device of claim 8 , wherein the data acquired from the plurality of cameras includes images or videos that capture the ball, wherein a ball-throwing machine has imparted motion to the ball.
12. The computing device of claim 8 , wherein computing the three-dimensional topography from the identified image includes utilization of a triangulation technique to determine a position of the ball and a position of the human in three-dimensional space.
13. The computing device of claim 8 , wherein extracting the features for both the ball and the human in each of the plurality of images that were not rejected from further processing includes:
recreating a three-dimensional relief of the user and of the ball, and
extracting pixel-based features of the ball and of the human from the three-dimensional relief, wherein the pixel-based features extracted from the three-dimensional relief are used in determining the greatest distance between the ball and the human.
14. The computing device of claim 8 , further comprising:
determining an image to be displayed to the human based on the three-dimensional topography computed from the identified image.
15. A non-transitory storage medium having stored thereon instructions that, when executed, cause performance of operations including:
acquiring data from a plurality of cameras surrounding a training environment, the data including a plurality of images,
attempting to segment a ball and a human in each of the plurality of images, wherein responsive to failing to segment either the ball or the human in a first image, rejecting the first image from further processing,
creating a mask of the ball and a mask of the human,
extracting features for both the ball and the human in each of the plurality of images that were not rejected from further processing,
computing surface information of the training environment using a photoclinometry process,
for each of the plurality of images that were not rejected from further processing, computing a distance between the ball and the human,
identifying an image of the plurality of images having a greatest distance between the ball and the human, and
computing a three-dimensional topography from the identified image.
16. The non-transitory storage medium of claim 15 , wherein the distance between the ball and the human is a Euclidean distance.
17. The non-transitory storage medium of claim 15 , wherein the photoclinometry process includes operations that transform a two-dimensional (2D) image into a surface map representing varying depth within the 2D image.
18. The non-transitory storage medium of claim 15 , wherein the data acquired from the plurality of cameras includes images or videos that capture the ball, wherein a ball-throwing machine has imparted motion to the ball.
19. The non-transitory storage medium of claim 15 , wherein computing the three-dimensional topography from the identified image includes utilization of a triangulation technique to determine a position of the ball and a position of the human in three-dimensional space.
20. The non-transitory storage medium of claim 15 , wherein extracting the features for both the ball and the human in each of the plurality of images that were not rejected from further processing includes:
recreating a three-dimensional relief of the user and of the ball, and
extracting pixel-based features of the ball and of the human from the three-dimensional relief, wherein the pixel-based features extracted from the three-dimensional relief are used in determining the greatest distance between the ball and the human.