Real-time detection of object scanability
A system and method are disclosed for determining and alerting a user as to whether an object will successfully scan before the post-processing of the scan data. In embodiments, before post-processing of the scan data begins, the scan data is processed by a machine learning algorithm which is able to determine whether and/or how likely the scan data is to return an accurate scanned reproduction of the scanned object. The machine learning algorithm may also suggest new positions for the object in the environment where the scan is more likely to be successful.
1. A method, comprising:
receiving scan data from scans of an object in a plurality of trials;
post-processing the scan data to generate a scanned reproduction of the object based on the scan data;
determining, using the scanned reproduction of the object, objective outputs in the plurality of trials as to whether the post-processing of the scan data can return an accurate scanned reproduction of the object in the plurality of trials;
training a machine learning algorithm over the plurality of trials to improve a correspondence between the objective outputs and a calculated output of the machine learning algorithm for the scan data, the machine learning algorithm operable to predict, upon completion of a new scan of a new object and prior to post-processing new scan data produced by the new scan, whether the new scan data can return an accurate scanned reproduction of the new object;
applying the machine learning algorithm to the new scan of the new object to determine whether the new scan of the new object can return the accurate scanned reproduction of the new object; and
suggesting an alternative position for the new object based, at least in part, on scan environment attributes included in the new scan data, in response to determining, prior to the post-processing of the new scan data, that the new scan is unlikely to return the accurate scanned reproduction of the new object.
2. The method of claim 1 , further comprising:
receiving feedback on an output of the applying of the machine learning algorithm to the new scan of the new object; and
updating the machine learning algorithm based on the feedback.
3. The method of claim 1 , wherein the training of the machine learning algorithm over the plurality of trials comprises refining numerical weights used by the machine learning algorithm to improve the correspondence between the objective outputs and the calculated output.
4. The method of claim 1 , wherein the machine learning algorithm is operable to predict, in real time, upon completion of the new scan of the new object and prior to the post-processing of the new scan data produced by the new scan, whether the new scan data can return the accurate scanned reproduction of the new object.
5. A computing device, comprising:
one or more scanners comprising one or more image sensors operable to produce scan data from scans of an object in a plurality of trials; and
a processor operable to:
receive the scan data from the one or more image sensors;
post-process the scan data to generate a scanned reproduction of the object based on the scan data;
determine, using the scanned reproduction of the object, objective outputs in the plurality of trials as to whether the post-processing of the scan data can return an accurate scanned reproduction of the object in the plurality of trials;
train a machine learning algorithm over the plurality of trials to improve a correspondence between the objective outputs and a calculated output of the machine learning algorithm for the scan data, the machine learning algorithm operable to predict, upon completion of a new scan of a new object and prior to post-processing new scan data produced by the new scan, whether the new scan data can return an accurate scanned reproduction of the new object;
apply the machine learning algorithm to the new scan of the new object to determine whether the new scan of the new object can return the accurate scanned reproduction of the new object; and
suggest an alternative position for the new object based, at least in part, on scan environment attributes included in the new scan data, in response to determining, prior to the post-processing of the new scan data, that the new scan is unlikely to return the accurate scanned reproduction of the new object.
6. The computing device of claim 5 , wherein the machine learning algorithm is operable to predict, in real time, upon completion of the new scan of the new object and prior to the post-processing of the new scan data produced by the new scan, whether the new scan data can return the accurate scanned reproduction of the new object.
7. The computing device of claim 5 , wherein the processor is further operable to notify a user that the new scan is unlikely to return the accurate scanned reproduction of the new object, in response to determining, prior to the post-processing of the new scan data, that the new scan is unlikely to return the accurate scanned reproduction of the new object.
8. The computing device of claim 5 , wherein the processor is further operable to perform the post-processing of the scan data in each trial of the plurality of trials, to reach a number of objective outputs, one respective objective output in each trial, as to whether the post-processing of the scan data received for each trial can return a respective accurate reproduction of the object.
9. The computing device of claim 5 , wherein the processor is further operable to suggest the alternative position for the new object by:
defining a volume where the new scan takes place; and
synthetically re-injecting the volume onto another surface.
10. The computing device of claim 5 , wherein the machine learning algorithm comprises an artificial neural network including numeric weights.
11. The computing device of claim 10 , wherein the training of the machine learning algorithm over the plurality of trials comprises refining the numerical weights to improve the correspondence between the objective outputs and the calculated output.
12. The computing device of claim 5 , wherein the processor is further operable to:
receive feedback on an output of the applying of the machine learning algorithm to the new scan of the new object, and
update the machine learning algorithm based on the feedback.
13. The computing device of claim 5 , wherein the computing device is used in association with a head mounted display device providing an augmented reality experience comprising the scanned reproduction of the new object.
14. A computer readable medium for storing computer instructions executed by one or more processors to perform the steps of:
receiving scan data from scans of an object in a plurality of trials;
post-processing the scan data to generate a scanned reproduction of the object based on the scan data;
determining, using the scanned reproduction of the object, objective outputs in the plurality of trials as to whether the post-processing of the scan data can return an accurate scanned reproduction of the object in the plurality of trials;
training a machine learning algorithm over the plurality of trials to improve a correspondence between the objective outputs and a calculated output of the machine learning algorithm for the scan data, the machine learning algorithm operable to predict, upon completion of a new scan of a new object and prior to post-processing new scan data produced by the new scan, whether the new scan data can return an accurate scanned reproduction of the new object;
applying the machine learning algorithm to the new scan of the new object to determine whether the new scan of the new object can return the accurate scanned reproduction of the new object, and
suggesting an alternative position for the new object based, at least in part, on scan environment attributes included in the new scan data, in response to determining, prior to the post-processing of the new scan data, that the new scan is unlikely to return the accurate scanned reproduction of the new object.
15. The computer readable medium of claim 14 , wherein the new scan data comprises one or more data streams received from one or more image sensors of a head mounted display.
16. The computer readable medium of claim 15 , wherein the one or more data streams include one or more of depth data, RGB data, infrared data, or pose data.
17. The computer readable medium of claim 14 , wherein the instructions when executed by the one or more processors cause the one or more processors to perform further steps of:
receiving feedback on an output of the applying of the machine learning algorithm to the new scan of the new object; and
updating the machine learning algorithm based on the feedback.
18. The computer readable medium of claim 14 , wherein the instructions when executed by the one or more processors cause the one or more processors to perform further steps of:
receiving an indication to abort the post processing of the new scan data, in response to applying the machine learning algorithm to the new scan of the new object to determine that the new scan of the new object cannot return the accurate scanned reproduction of the new object.