Confidence generation using a neural network
Apparatuses, systems, and techniques to generate one or more confidence values associated with one or more objects identified by one or more neural networks. In at least one embodiment, one or more confidence values associated with one or more objects identified by one or more neural networks are generated based on, for example, one or more neural network outputs.
1. One or more processors, comprising circuitry to use one or more neural networks to:
identify one or more objects in one or more images based, at least in part, on segmentation of the one or more images;
identify, based on the segmentation, one or more probabilities indicating that one or more pixels of the one or more objects is part of a corresponding object;
generate one or more confidence values indicating predicted classifications to be associated with the one or more objects based, at least in part, on two or more probabilities generated for a same pixel; and
in response to the one or more confidence values, cause one or more actions to be performed, wherein the one or more actions includes at least activating one or more steering systems of a vehicle.
2. The one or more processors of claim 1 , wherein the one or more neural networks comprise one or more image segmentation networks, and further wherein the one or more objects are identified as belonging to one or more segmentation classes.
3. The one or more processors of claim 1 , wherein the circuitry is to further use the one or more neural networks to generate one or more class scores corresponding to the one or more objects.
4. The one or more processors of claim 3 , wherein the circuitry is to further use the one or more neural networks to at least:
perform calibration on the one or more class scores; and
generate a label based at least in part on results from the calibration.
5. The one or more processors of claim 1 , wherein the circuitry is to further cause the vehicle to perform the one or more remedial actions in response to the one or more confidence values.
6. The one or more processors of claim 1 , wherein the one or more actions include activating one or more warning systems of the vehicle.
7. The one or more processors of claim 1 , wherein the one or more neural networks generate a confidence map that corresponds to a semantic segmentation of the one or more images.
8. The one or more processors of claim 7 , wherein the confidence map comprises a visualization of one or more confidence values based, at least in part, on a gradient value.
9. The one or more processors of claim 1 , wherein the one or more neural networks cause the vehicle to perform one or more actions based, at least in part, on one or more confidence values not satisfying a threshold value of confidence.
10. A system, comprising one or more processors to use one or more neural networks to:
identify one or more objects in one or more images based, at least in part, on segmentation of the one or more images;
identify, based on the segmentation, one or more probabilities indicating that one or more pixels of the one or more objects is part of a corresponding object;
generate one or more confidence values indicating predicted classifications to be associated with the one or more objects based, at least in part, on two or more probabilities generated for a same pixel; and
in response to the one or more confidence values, cause one or more actions to be performed, wherein the one or more actions includes at least activating one or more steering systems of a vehicle.
11. The system of claim 10 , wherein the one or more objects are identified from one or more images captured using one or more cameras of a vehicle.
12. The system of claim 10 , wherein the one or more outputs includes a vector of scores associated with one or more labels for each pixel of an image depicting the one or more objects.
13. The system of claim 12 , wherein the one or more processors are further to cause a semi-autonomous vehicle to perform one or more remedial actions in response to the vector of scores.
14. The system of claim 13 , wherein the one or more remedial actions include activating one or more acceleration systems of the semi-autonomous vehicle.
15. The system of claim 13 , wherein the one or more remedial actions include activating one or more braking systems of the semi-autonomous vehicle.
16. The system of claim 13 , wherein the one or more remedial actions include activating one or more warning systems of the semi-autonomous vehicle.
17. The system of claim 10 , wherein the one or more neural networks further adjusts one or more confidence values based, at least in part, on one or more environmental conditions detected by one or more sensors of the vehicle.
18. A non-transitory machine-readable medium having stored thereon a set of instructions, which if performed by one or more processors, cause the one or more processors to use one or more neural networks to:
identify one or more objects in one or more images based, at least in part, on segmentation of the one or more images;
identify, based on the segmentation, one or more probabilities indicating that one or more pixels of the one or more objects is part of a corresponding object;
generate one or more confidence values indicating predicted classifications to be associated with the one or more objects based, at least in part, on two or more probabilities generated for a same pixel; and
in response to the one or more confidence values, cause one or more actions to be performed, wherein the one or more actions include activating one or more medical warning systems.
19. The non-transitory machine-readable medium of claim 18 , wherein the one or more objects are identified from an image captured from one or more medical imaging devices.
20. The non-transitory machine-readable medium of claim 18 , wherein one or more confidence values indicate one or more probabilities of whether one or more identifications of the one or more objects are correct.
21. The non-transitory machine-readable medium of claim 18 , wherein the set of instructions further include instructions, which if performed by the one or more processors, cause the one or more processors to use the one or more neural networks to:
determine whether one or more confidence values are below a threshold; and
as a result of determining that the one or more confidence values are below the threshold, perform one or more actions.
22. The non-transitory machine-readable medium of claim 18 , wherein the one or more actions include transmitting a notification to one or more medical systems.
23. The non-transitory machine-readable medium of claim 18 , wherein the set of instructions further include instructions, which if performed by the one or more processors, cause the one or more processors to use the one or more neural networks to generate a segmentation map visualization indicating identifications of the one or more objects.
24. One or more processors, comprising circuitry to train one or more neural networks to:
identify one or more objects in one or more images based, at least in part, on segmentation of the one or more images;
identify, based on the segmentation, one or more probabilities indicating that one or more pixels of the one or more objects is part of a corresponding object;
generate one or more confidence values indicating predicted classifications to be associated with the one or more objects based, at least in part, on two or more probabilities generated for a same pixel; and
in response to the one or more confidence values, cause one or more actions to be performed, wherein the one or more actions include activating one or more warning systems.
25. The one or more processors of claim 24 , wherein one or more confidence values are generated from one or more hidden layers of the one or more neural networks and provided with inferencing output.
26. The one or more processors of claim 24 , wherein the one or more objects are identified from images captured by one or more geosensing imaging system, and further wherein an indication of multiple confidence values are transmitted to the one or more geosensing imaging systems.
27. The one or more processors of claim 24 , wherein the circuitry is further to cause the one or more images to be segmented into one or more types of objects based, at least in part, on a comparison of how likely the one or more objects belong to at least one of two different semantic classes.
28. The one or more processors of claim 27 , wherein the comparison is performed by the one or more neural networks based, at least in part, on the two or more probabilities for the same pixel indicating probabilities of different predicted classes.
29. The one or more processors of claim 27 , wherein the circuitry is to further perform one or more actions causing the one or more neural networks to generate a different set of confidence values associated with the one or more objects.
30. A system, comprising one or more processors to train one or more neural networks to:
identify one or more objects in one or more images based, at least in part, on segmentation of the one or more images;
identify, based on the segmentation, one or more probabilities indicating that one or more pixels of the one or more objects is part of a corresponding object;
generate one or more confidence values indicating predicted classifications to be associated with the one or more objects based, at least in part, on two or more probabilities generated for a same pixel; and
in response to the one or more confidence values, cause one or more actions to be performed, wherein the one or more actions include causing to at least activate one of:
one or more fertilization systems;
one or more watering systems; or
one or more weeding systems.
31. The system of claim 30 , wherein the one or more processors are further to generate one or more classification scores associated with the one or more objects.
32. The system of claim 31 , wherein the one or more processors are further to perform one or more softmax processes on the one or more classification scores.
33. The system of claim 30 , wherein the one or more objects are identified from one or more images obtained from one or more agriculture devices.
34. The system of claim 30 , wherein the one or more processors are further to train the one or more neural networks to generate one or more confidence values for at least one of: instance segmentation, panoptic segmentation, optical flow, stereo disparity, classification, or detection processes.