Using a probabilistic model for detecting an object in visual data

1. A method comprising:

performing, by a system having a processor and a non-transitory storage computer-readable medium:

receiving a given object;

generating simulated views of the given object;

obtaining matching statistics output from matching features of each simulated view of the simulated views of the given object to representations of objects stored in a database;

obtaining non-matching statistics output from matching features of each reference image of a plurality of reference images that does not include the given object to the representations of objects; and

building a probabilistic model, the probabilistic model comprising a conditional probability derived utilizing the matching statistics and non-matching statistics outputs of the matching the feature of the each simulated view of the simulated views of the given object to the representations of the objects and the matching the features of the each reference image of the plurality of reference images to the representations of the objects, wherein the probabilistic model is usable for detecting whether a particular object is present in received visual data.

2. The method of claim 1 , wherein building the probabilistic model comprises determining or obtaining a number of inliers, each of which representing a point correspondence between a point feature of an image and a point feature of a representation of an object stored in the database.

3. The method of claim 1 , further comprising:

providing the probabilistic model to another system communicatively connected to the system over a network.

4. The method of claim 1 , wherein the probabilistic model is further configured to produce an output of a number threshold and wherein the system stores the number threshold for later use by a remote system communicatively connected to the system over a network.

5. The method of claim 1 , wherein the simulated views correspond to different poses of a camera with respect to the given object.

6. The method of claim 1 , wherein using the probabilistic model comprises:

determining a number of matching features between the received visual data and a corresponding one of the representations of objects;

calculating a probability value by the probabilistic model in response to the determined number of matching features; and

comparing the probability value to a predetermined probability threshold.

7. The method of claim 1 , wherein using the probabilistic model comprises:

determining a number of matching features between the received visual data and a corresponding one of the representations of objects;

calculating a number threshold by the probabilistic model in response to a predefined probability threshold; and

comparing the determined number of matching features to the number threshold.

8. The method of claim 1 further comprising:

providing an additional probabilistic model for an additional object; and

using the additional probabilistic model for detecting whether the additional object is present in received visual data.

9. The method of claim 8 , wherein using the probabilistic models for detecting the objects are present in respective visual data employs a probability threshold.

10. An article comprising at least one processor and at least one non-transitory machine-readable storage medium storing instructions that, upon execution by the at least one processor, cause a system to:

receive a given object;

generate simulated views of the given object;

obtain matching statistics output from matching features of each simulated view of the simulated views of the given object to representations of objects stored in a database accessible by the system;

obtain non-matching statistics output from matching features of each reference image of a plurality of reference images that does not include the given object to the representations of objects; and

build a probabilistic model, the probabilistic model comprising a conditional probability derived utilizing the matching statistics and non-matching statistics outputs of the matching the features of the each simulated view of the simulated views of the given object to the representations of the objects and the matching the features of the each reference image of the plurality of reference images to the representations of the objects, wherein the probabilistic model is useable to detect whether an object is in visual data.

11. The article of claim 10 , wherein the simulated views corresponding to different poses of a camera with respect to the given object.

12. The article of claim 10 , wherein the conditional probability further comprises a probability that indicates whether the given object is in visual data, given a number of matching features.

13. The article of claim 10 , wherein the instructions upon execution by the at least one processor further cause the system to:

receive the visual data;

determine a number of matching features between the visual data and a given one of the representations of the objects; and

use the probabilistic model and the determined number of matching features to determine whether the object represented by the given representation is present in the visual data.

14. The article of claim 13 , wherein using the probabilistic model is based on a probability threshold.

15. A system comprising:

at least one processor;

non-transitory machine-readable storage medium; and

stored instructions translatable by the at least one processor to:

receive a given object;

generate simulated views of the given object;

obtain matching statistics output from matching features of each simulated view of the simulated views of the given object to representations of objects stored in a database;

obtain non-matching statistics output from matching features of each reference image of a plurality of reference images that does not include the given object to the representations of objects; and

build a probabilistic model, the probabilistic model comprising a conditional probability derived utilizing the matching statistics and non-matching statistics outputs of the matching the features of the each simulated view of the simulated views of the given object to the representations of the objects and the matching the features of the each reference image of the plurality of reference images to the representations of the objects, wherein the probabilistic model is usable for detecting whether a particular object is present in received visual data.

16. The system of claim 15 , wherein the system is communicatively connected to a remote system over a network, wherein the system is operable to provide the probabilistic model to the remote system over the network, and wherein the probabilistic model, which is built by the system, is usable by the remote system.

17. The system of claim 15 , wherein the simulated views correspond to different poses of a camera with respect to the given object.

18. The system of claim 15 , wherein the conditional probability further comprises a probability that indicates whether the given object is in visual data, given a number of matching features.

19. The system of claim 15 , wherein the probabilistic model is built for a single object type and wherein the system is operable to build different probabilistic models for different object types, utilizing the statistics and non-matching statistics outputs.

20. The system of claim 15 , wherein the probabilistic model is built for different object types, the probabilistic model configured for performing different conversions between a probability value and a respective number of point correspondences for the different object types, each point correspondence representing a match or mismatch between a feature of an image and a feature of a representation of an object.