System and method for neural network orchestration
Methods and systems for classifying a multimedia file using interclass data is disclosed. One of the methods includes receiving, from a first transcription engine, one or more transcription results of one or more audio segments of the multimedia file; identifying a first transcription result for a first audio segment having a low confidence of accuracy; identifying a first image data of the multimedia file corresponding to the first segment; receiving, from an image classification engine trained to classify image data, an image classification result of one or more portions of the first image data in response to requesting the image classification engine to classify the first image data; and selecting, based at least on the image classification result of the one or more portions of the first image data, a second transcription engine to re-classify the first audio segment.
1. A method for classifying a media file having two or more types of data, the method comprising:
receiving, from a first classification neural network, one or more classification results of a first group of one or more segments of the media file, wherein the first group of one or more segments comprises a first type of data;
identifying a first classification result of a first segment with a low confidence of accuracy from the one or more classification results of the first group;
identifying a second group of one or more segments of the media file related to the first segment, wherein the second group of one or more segments comprises a second type of data, and wherein the first and second types of data are different;
receiving, from a second classification neural network trained to classify the second type of data, a second classification result of the second group of one or more segments in response to requesting the second classification neural network to classify the second group of one or more segments; and
selecting, based at least on the second classification result, a third classification neural network to re-classify the first segment with a low confidence of accuracy, wherein the first and third classification neural networks are trained to classify the first type of data.
2. The method of claim 1 , wherein the first type comprises audio data, and the second type comprises image data or metadata.
3. The method of claim 2 , wherein the first and third classification neural networks are trained to perform speech-to-text transcription using the audio data of the media file, and wherein the second classification neural network is trained to perform image classification.
4. The method of claim 2 , wherein the first and third classification neural networks are trained to perform speech-to-text transcription using the audio data of the media file, and wherein the second classification neural network is trained to perform topic classification, and wherein the second type of data comprises metadata.
5. The method of claim 1 , wherein the first type comprises image data, and the second type comprises audio data or metadata.
6. The method of claim 5 , wherein the first and third classification neural networks are trained to perform object recognition using the image data of the media file, and wherein the second classification neural network is trained to perform speech-to-text transcription using the audio data of the media file.
7. The method of claim 1 , wherein the third classification neural network is configured to re-classify the first segment based at least on the second classification result.
8. The method of claim 1 , wherein identifying the second group of one or more segments of the media file related to the first segment comprises identifying one or more segments of the second group that spans ±10 seconds with respect to a time position of the first segment within the media file.
9. The method of claim 8 , wherein the one or more segments of the second group have approximately a same time duration within the media file as the first segment.
10. The method of claim 1 , further comprising:
receiving a third classification result from the third classification neural network;
replacing the first classification result with the third classification result if the confidence of accuracy of the third classification result is higher than a confidence of accuracy of the first classification result.
11. A system for classifying a media file having two or more types of data, the system comprising:
a memory; and
one or more processors coupled to the memory, the one or more processor configured to:
receive, from a first classification neural network, one or more classification results of a first group of one or more segments of the media file, wherein the first group of one or more segments comprises a first type of data;
identify a first classification result of a first segment with a low confidence of accuracy from the one or more classification results of the first group;
identify a second group of one or more segments of the media file related to the first segment, wherein the second group of one or more segments comprises a second type of data, and wherein the first and second types of data are different;
receive, from a second classification neural network trained to classify the second type of data, a second classification result of the second group of one or more segments in response to requesting the second classification neural network to classify the second group of one or more segments; and
select, based at least on the second classification result, a third classification neural network to re-classify the first segment with a low confidence of accuracy, wherein the first and third classification neural networks are trained to classify the first type of data.
12. The system of claim 11 , wherein the first type comprises audio data, and the second type comprises image data or metadata.
13. The system of claim 12 , wherein the first and third classification neural networks are trained to perform speech-to-text transcription using the audio data of the media file, and wherein the second classification neural network is trained to perform image classification.
14. The system of claim 11 , wherein the first type comprises image data, and the second type comprises audio data or metadata.
15. The system of claim 14 , wherein the first and third classification neural networks are trained to perform object recognition using the image data of the media file, and wherein the second classification neural network is trained to perform speech-to-text transcription using the audio data of the media file.
16. The system of claim 11 , wherein the third classification neural network is configured to re-classify the first segment based at least on the second classification result.
17. The system of claim 11 , wherein identifying the second group of one or more segments of the media file related to the first segment comprises identifying one or more segments of the second group that spans ±10 seconds with respect to a time position of the first segment within the media file.
18. A method for transcribing a multimedia file, the method comprising:
receiving, from a first transcription engine, one or more transcription results of one or more audio segments of the multimedia file;
identifying a first transcription result for a first audio segment having a low confidence of accuracy;
identifying a first image data of the multimedia file corresponding to the first segment;
receiving, from an image classification engine trained to classify image data, an image classification result of one or more portions of the first image data in response to requesting the image classification engine to classify the first image data; and
selecting, based at least on the image classification result of the one or more portions of the first image data, a second transcription engine to re-classify the first audio segment, wherein the first and second transcription engines are different.
19. The method of claim 18 , identifying the first image data of the multimedia file corresponding to the first segment comprises identifying image data that spans ±10 seconds with respect to a time position of the first audio segment within the multimedia file.
20. The method of claim 18 , identifying the first image data of the multimedia file corresponding to the first segment comprises identifying image data that occur at approximately at a same time as the first audio segment within the multimedia file.
21. A method for classifying a media segment of a first data type having a corresponding media segment of a second data type, the method comprising:
extracting a first set of media features from the media segment of the first data type;
generating, using an engine prediction neural network, a best candidate neural network based on the first set of media features, wherein the best candidate neural network comprises a neural network having a highest predicted value of accuracy;
determining whether a predicted value of accuracy of the best candidate neural network is above a predetermined accuracy threshold;
when the predicted value of accuracy of the best candidate neural network is below the predetermined accuracy threshold, classifying the corresponding media segment of a second data type using a second classification neural network; and
selecting, based at least on results of the classification of the corresponding media segment of a second data type, a third classification neural network to classify the media segment of the first data type, wherein the first and second data types are different, and wherein the third classification neural network and the best candidate neural network are different.
22. The method of claim 21 , wherein extracting the first set of media features of the first data type comprises extracting audio features of the media segment using outputs of one or more layers of a speech-to-text classification neural network, wherein the first data type comprises audio data and the second data type comprises image data or metadata.
23. The method of claim 21 , wherein extracting the first set of media features of the first data type comprises extracting image features of the media segment using outputs of one or more layers of an image classification neural network, wherein the first data type comprises image data and the second data type comprises audio data or metadata.
24. The method of claim 21 , further classifying the corresponding media segment of a second data type using the second classification neural network comprising:
extracting a second set of media features from the corresponding media segment of the second data type; and
generating, using the engine prediction neural network, a best candidate neural network based on the second set of media features, wherein the second classification neural network comprises the best candidate neural network.
25. A method for classifying a portion of an image, the method comprising:
receiving, from a first image classification engine, an image classification results of the portion of the image;
determining a confidence of accuracy value of the image classification results received from the first image classification engine;
identifying an alternate data set associated with the image, the alternate data set comprises non-image data;
requesting a second classification engine to classify the alternate data set, wherein the second classification engine is trained to classify data in a same class as the alternate data set;
receiving, from the second classification engine, a second classification result of the alternate data set; and
selecting, based at least on the second classification result of the alternate data set, a third image classification engine to re-classify the portion of the image, wherein the first and third image classification engines are different.
26. The method of claim 25 , wherein the alternate data set comprises audio data occurring within 30 seconds before and after from an instant the image appearing in the multimedia file.
27. The method of claim 25 , wherein the alternate data set comprises metadata occurring within 1 minute before and after from an instant the image appearing in the multimedia file.
28. The method of claim 25 , wherein the alternate data set comprises transcription occurring within 1 minute before and after from an instant the image appearing in the multimedia file.