Machine learning in video classification
Described herein are systems and methods that search videos and other media content to identify items, objects, faces, or other entities within the media content. Detectors identify objects within media content by, for instance, detecting a predetermined set of visual features corresponding to the objects. Detectors configured to identify an object can be trained using a machine learned model (e.g., a convolutional neural network) as applied to a set of example media content items that include the object. The systems provide user interfaces that allow users to review search results, pinpoint relevant portions of media content items where the identified objects are determined to be present, review detector performance and retrain detectors, providing search result feedback, and/or reviewing video monitoring results and analytics.
1. A computer-implemented method, comprising:
receiving, from a user, a search query identifying an object;
classifying frames of a video using one or more detectors configured to, for each frame of the video, process the frame and output a confidence score indicating a likelihood that the identified object is present within the frame;
identifying a first set of video segments and a second set of video segments based on the classification, each video segment in the first set of video segments comprising a set of consecutive frames associated with confidence scores greater than a first threshold confidence score and each video segment in the second set of video segments comprising a set of consecutive frames associated with confidence scores greater than a second threshold confidence score; and
modifying a progress bar interface element comprising a first color to display highlighted locations within the progress bar interface element that correspond to the first set of video segments in a second color different from the first color and that correspond to the second set of video segments in a third color different from the first and second colors.
2. The method of claim 1 , wherein the one or more detectors comprise a plurality of detectors, and wherein the confidence score associated with the frame classified by the plurality of detectors comprises an average of confidence scores outputted by the plurality of detectors in response to processing the frame.
3. The method of claim 1 , wherein each video segment in the first set of video segments and the second set of video segments comprises at least a threshold number of consecutive frames.
4. The method of claim 1 , wherein an average confidence score of the set of consecutive frames of a video segment in the first set of video segments is greater than the first threshold confidence score, and wherein at least one confidence score associated with a frame of the set of consecutive frames of the video segment is below the first threshold confidence score.
5. The method of claim 1 , wherein the first threshold confidence score and the second threshold confidence score are received from the user.
6. The method of claim 1 , further comprising:
monitoring a displayed frame of the video during playback of the video; and
providing the confidence score associated with the displayed frame for display.
7. A non-transitory computer-readable storage medium storing executable instructions that, when executed by a hardware processor, cause the hardware processor to perform steps comprising:
receiving, from a user, a search query identifying an object;
classifying frames of a video using one or more detectors configured to, for each frame of the video, process the frame and output a confidence score indicating a likelihood that the identified object is present within the frame;
identifying a first set of video segments and a second set of video segments based on the classification, each video segment in the first set of video segments comprising a set of consecutive frames associated with confidence scores greater than a first threshold confidence score and each video segment in the second set of video segments comprising a set of consecutive frames associated with confidence scores greater than a second threshold confidence score; and
modifying a progress bar interface element comprising a first color to display highlighted locations within the progress bar interface element that correspond to the first set of video segments in a second color different from the first color and that correspond to the second set of video segments in a third color different from the first and second colors.
8. The non-transitory computer-readable storage medium of claim 7 , wherein the one or more detectors comprise a plurality of detectors, and wherein the confidence score associated with the frame classified by the plurality of detectors comprises an average of confidence scores outputted by the plurality of detectors in response to processing the frame.
9. The non-transitory computer-readable storage medium of claim 7 , wherein each video segment in the first set of video segments and the second set of video segments comprises at least a threshold number of consecutive frames.
10. The non-transitory computer-readable storage medium of claim 7 , wherein an average confidence score of the set of consecutive frames of a video segment in the first set of video segments is greater than the first threshold confidence score, and wherein at least one confidence score associated with a frame of the set of consecutive frames of the video segment is below the first threshold confidence score.
11. The non-transitory computer-readable storage medium of claim 7 , wherein the first threshold confidence score and the second threshold confidence score are received from the user.
12. The non-transitory computer-readable storage medium of claim 7 , wherein the instructions, when executed by the hardware processor, cause the hardware processor to perform further steps comprising:
monitoring a displayed frame of the video during playback of the video; and
providing the confidence score associated with the displayed frame for display.
13. A system, comprising:
a computer processor; and
a non-transitory memory storing executable computer instructions that when executed by the computer processor are configured to cause the computer processor to perform steps comprising:
receiving, from a user, a search query identifying an object;
classifying frames of a video using one or more detectors configured to, for each frame of the video, process the frame and output a confidence score indicating a likelihood that the identified object is present within the frame;
identifying a first set of video segments and a second set of video segments based on the classification, each video segment in the first set of video segments comprising a set of consecutive frames associated with confidence scores greater than a first threshold confidence score and each video segment in the second set of video segments comprising a set of consecutive frames associated with confidence scores greater than a second threshold confidence score; and
modifying a progress bar interface element comprising a first color to display highlighted locations within the progress bar interface element that correspond to the first set of video segments in a second color different from the first color and that correspond to the second set of video segments in a third color different from the first and second colors.
14. The system of claim 13 , wherein the one or more detectors comprise a plurality of detectors, and wherein the confidence score associated with the frame classified by the plurality of detectors comprises an average of confidence scores outputted by the plurality of detectors in response to processing the frame.
15. The system of claim 13 , wherein each video segment in the first set of video segments and the second set of video segments comprises at least a threshold number of consecutive frames.
16. The system of claim 13 , wherein an average confidence score of the set of consecutive frames of a video segment in the first set of video segments is greater than the first threshold confidence score, and wherein at least one confidence score associated with a frame of the set of consecutive frames of the video segment is below the first threshold confidence score.
17. The system of claim 13 , wherein the first threshold confidence score and the second threshold confidence score are received from the user.