Prediction model training via live stream concept association
In certain embodiments, training of a neural network or other prediction model may be facilitated via live stream concept association. In some embodiments, a live video stream may be loaded on a user interface for presentation to a user. A user selection related to a frame of the live video stream may be received via the user interface during the presentation of the live video stream on the user interface, where the user selection indicates a presence of a concept in the frame of the live video stream. In response to the user selection related to the frame, an association of at least a portion of the frame of the live video stream and the concept may be generated, and the neural network or other prediction model may be trained based on the association of at least the portion of the frame with the concept.
1. A system that comprises one or more processors programmed with computer program instructions that, when executed, cause the system to:
display, on a user interface, a video stream and a plurality of predicted concepts associated with the video stream, the plurality of predicted concepts generated by a prediction model based on the video stream, and the plurality of predicted concepts include a first set of predicted concepts and a second set of predicted concepts, the first set of predicted concepts being associated with first confidence values equal to or higher than a confidence threshold and the second set of predicted concepts being associated with second confidence values less than the confidence threshold, and the first and second confidence values indicate a confidence level that a particular predicted concept is present in a particular frame of the video stream;
receive, from an input device, a user selection of a concept related to a frame of the video stream, the selected concept being from among the plurality of predicted concepts;
receive, from the input device, a selection of a portion of the frame of the video stream;
determine an association between the selected portion of the frame and the selected concept; and
cause the prediction model to be trained based on the association.
2. The system of claim 1 further comprising instructions that, when executed, cause the system to:
determine a positive association between the selected portion of the frame and the selected concept, the positive association indicating the presence of the selected concept in the selected portion of the frame; and
store the positive association in a memory;
wherein the prediction model is trained based on the stored positive association.
3. The system of claim 1 further comprising instructions that, when executed, cause the system to:
determine a negative association between the selected portion of the frame and the selected concept, the negative association indicating the absence of the selected concept in the selected portion of the frame; and
store the negative association in a memory;
wherein the prediction model is trained based on the stored negative association.
4. The system of claim 2 , further comprising instructions that, when executed, cause the system to:
receive, from the input device, a user selection of a second concept related to a second frame of the video stream, the selection of the second concept being from among the plurality of predicted concepts;
receive, from the input device, a selection of a second portion of the second frame and the selected second concept.
5. The system of claim 4 , further comprising instructions that, when executed, cause the system to:
determine a negative association between the selected second portion of the second frame and the selected second concept, the negative association indicating the absence of the selected second concept in the selected second portion of the second frame; and
store the negative association in a memory;
wherein the prediction model is trained based on the stored negative association.
6. The system of claim 5 , wherein the selected concept and the selected second concept are the same concept from the plurality of predicted concepts.
7. The system of claim 1 , wherein receiving the user selection of the concept causes computer program instructions that, when executed, to further cause the system to:
determine a pressure level applied by the user on the user interface; and
determine a confidence value associated with the concept and the frame based on the determined pressure level,
wherein the prediction is trained based on the determined confidence value.
8. The system of claim 7 , wherein the confidence value is directly proportional to the pressure level.
9. The system of claim 1 , wherein the input device is one of a touch screen, a microphone, a keyboard, and a mouse.
10. The system of claim 1 , wherein the first set of predicted concepts and the second set of predicted concepts are presented via the user interface such that the first set of predicted concepts and the second set of predicted concepts are distinguished from each other.
11. A computer-implemented method, comprising:
displaying, on a user interface, a video stream and a plurality of predicted concepts associated with the video stream, the plurality of predicted concepts generated by a prediction model based on the video stream, and the plurality of predicted concepts include a first set of predicted concepts and a second set of predicted concepts, the first set of predicted concepts being associated with first confidence values equal to or higher than a confidence threshold and the second set of predicted concepts being associated with second confidence values less than the confidence threshold, and the first and second confidence values indicate a confidence level that a particular predicted concept is present in a particular frame of the video stream;
receiving, from an input device, a user selection of a concept related to a frame of the video stream, the selected concept being from among the plurality of predicted concepts;
receiving, from the input device, a selection of a portion of the frame of the video stream;
determining an association between the selected portion of the frame and the selected concept; and
causing the prediction model to be trained based on the association.
12. The computer-implemented method of claim 11 , further comprising:
determining a positive association between the selected portion of the frame and the selected concept, the positive association indicating the presence of the selected concept in the selected portion of the frame; and
storing the positive association in a memory;
wherein the prediction model is trained based on the stored positive association.
13. The computer-implemented method of claim 11 , further comprising:
determining a negative association between the selected portion of the frame and the selected concept, the negative association indicating the absence of the selected concept in the selected portion of the frame; and
storing the negative association in a memory;
wherein the prediction model is trained based on the stored negative association.
14. The computer-implemented method of claim 12 , further comprising:
receiving, from the input device, a user selection of a second concept related to a second frame of the video stream, the selection of the second concept being from among the plurality of predicted concepts;
receiving, from the input device, a selection of a second portion of the second frame and the selected second concept.
15. The computer-implemented method of claim 14 , further comprising:
determining a negative association between the selected second portion of the second frame and the selected second concept, the negative association indicating the absence of the selected second concept in the selected second portion of the second frame; and
storing the negative association in a memory;
wherein the prediction model is trained based on the stored negative association.
16. The computer-implemented method of claim 15 , wherein the selected concept and the selected second concept are the same concept from the plurality of predicted concepts.
17. The computer-implemented method of claim 11 , wherein receiving the user selection of the concept further comprises:
determining a pressure level applied by the user on the user interface; and
determining a confidence value associated with the concept and the frame based on the determined pressure level,
wherein the prediction is trained based on the determined confidence value.
18. A non-transitory computer-readable medium comprising instructions which when executed by a processor cause a computer to perform a method comprising:
displaying, on a user interface, a video stream and a plurality of predicted concepts associated with the video stream, the plurality of predicted concepts generated by a prediction model based on the video stream, and the plurality of predicted concepts include a first set of predicted concepts and a second set of predicted concepts, the first set of predicted concepts being associated with first confidence values equal to or higher than a confidence threshold and the second set of predicted concepts being associated with second confidence values less than the confidence threshold, and the first and second confidence values indicate a confidence level that a particular predicted concept is present in a particular frame of the video stream;
receiving, from an input device, a user selection of a concept related to a frame of the video stream, the selected concept being from among the plurality of predicted concepts;
receiving, from the input device, a selection of a portion of the frame of the video stream;
determining an association between the selected portion of the frame and the selected concept; and
causing the prediction model to be trained based on the association.
19. The non-transitory computer-readable medium of claim 18 , wherein the input device is one of a touch screen, a microphone, a keyboard, and a mouse.