Systems and methods for pitch determination
Systems and methods are provided for pitch determination. An example method includes obtaining an image depicting a structure, the image being captured via a user device positioned proximate to the structure. The image is segmented to identify, at least, a roof facet of the structure. An eave vector and a rake vector which are associated with the roof facet are determined. A normal vector of the roof facet is calculated based on the eave vector and the rake vector, and compared to a vector indicating a vertical direction such as gravity. The angle made out by the normal and a gravity vector may be utilized to calculate the pitch of the roof facet.
1. A method implemented by a system of one or more computers, the method comprising:
providing access to an image depicting a structure, the image being captured via a user device positioned proximate to the structure, the structure having a plurality of planar elements comprising at least a roof facet and one or more walls;
providing the image as input to a neural network, wherein the neural network outputs, at least, a surface normal associated with the roof facet and a surface normal associated with a particular wall of the one or more walls;
adjusting each surface normal based on a transform, wherein the transform adjusts at least the surface normal associated with the particular wall to be substantially orthogonal to a vertical orientation; and
extracting a pitch of the roof facet based on the adjusted surface normal associated with the roof facet and a gravity vector.
2. The method of claim 1 , wherein the neural network comprises a convolutional neural network trained to segment the planar elements into at least the roof facet or the wall.
3. The method of claim 2 , wherein the neural network further comprises one or more fully-connected layers which receive output from the convolutional neural network, and wherein the fully-connected layers are trained to output individual surface normals associated with individual planar elements.
4. The method of claim 1 , wherein the image was captured below a height of the structure.
5. The method of claim 1 , further comprising determining a plurality of surface normals corresponding to a plurality of roof facets.
6. The method of claim 1 , wherein the gravity vector is a unit vector oriented along the vertical orientation.
7. The method of claim 1 , further comprising determining the gravity vector based on a feature associated with the particular wall.
8. The method of claim 1 , wherein the one or more walls include a plurality of walls, and wherein the neural network outputs a surface normal associated with each of the walls.
9. The method of claim 8 , wherein the transform adjusts the surface normal associated with each of the walls to be substantially orthogonal to vertical.
10. The method of claim 8 , wherein the transform adjusts individual surface normals associated with a subset of the walls to be substantially orthogonal to vertical.
11. A system comprising one or more processors and non-transitory computer readable media storing instructions which, when executed by the one or more processors, cause the one or more processors to:
provide access to an image depicting a structure, the image being captured via a user device positioned proximate to the structure, the structure having a plurality of planar elements comprising at least a roof facet and one or more walls;
provide the image as input to a neural network, wherein the neural network outputs, at least, a surface normal associated with the roof facet and a surface normal associated with a particular wall of the one or more walls;
adjust each surface normal based on a transform, wherein the transform adjusts at least the surface normal associated with the particular wall to be substantially orthogonal to a vertical orientation; and
extracting a pitch of the roof facet based on the adjusted surface normal associated with the roof facet and a gravity vector.
12. The system of claim 11 , wherein the neural network comprises a convolutional neural network trained to segment the planar elements into at least the roof facet or the wall.
13. The system of claim 12 , wherein the neural network further comprises one or more fully-connected layers which receive output from the convolutional neural network, and wherein the fully-connected layers are trained to output individual surface normals associated with individual planar elements.
14. The system of claim 11 , wherein the image was captured below a height of the structure.
15. The system of claim 11 , wherein the instructions further cause the one or more processors to determine a plurality of surface normals corresponding to a plurality of roof facets.
16. The system of claim 11 , wherein the gravity vector is a unit vector oriented along the vertical orientation.
17. The system of claim 11 , wherein the gravity vector is based on a feature associated with the particular wall.
18. The system of claim 11 , wherein the one or more walls include a plurality of walls, and wherein the neural network outputs a surface normal associated with each of the walls.
19. The system of claim 18 , wherein the transform adjusts the surface normal associated with each of the walls to be substantially orthogonal to vertical.
20. The system of claim 18 , wherein the transform adjusts individual surface normals associated with a subset of the walls to be substantially orthogonal to vertical.