Systems and methods for deep recommendations using signature analysis
Systems and methods are described herein for providing content item recommendations based on a video. Using feature vectors corresponding to at least one frame of a video (e.g., generated based on texture and shape intensity of a frame), a recommendation system improves content recommendation using analytic and quantitative characteristics derived from a frame of a content item rather than merely manually labeled bibliographic data (e.g., a genre or producer). The recommendation system may generate a feature vector based on a texture, a shape intensity (e.g., generated from a Generalized Hough Transform), and temporal data corresponding to at least one frame of a video. The feature vector is analyzed using a machine learning model (e.g., a neural network) to produce a machine learning model output. The recommendation system causes a recommended content item to be provided based on the machine learning model output.
1. A method comprising:
generating, using control circuitry, a first feature vector from at least one frame of a first content item, wherein generating the first feature vector comprises transforming the at least one frame of the first content item;
analyzing, using the control circuitry, the first feature vector using a machine learning model to produce a first machine learning model output;
identifying, based on the first machine learning model output, a second content item comprising at least one frame that is comprised of attributes corresponding to the first feature vector;
generating a second feature vector from at least one frame of the second content item, wherein generating the second feature vector comprises transforming the at least one frame of the second content item;
analyzing, using the control circuitry the second feature vector using the machine learning model to produce a second machine learning model output; and
identifying a recommended content item based on the first machine learning model output and the second machine learning model output;
determining whether the recommended content item has been generated as a recommendation previously;
determining, based on interaction data, whether there has been a previous interaction with the recommended content item;
in response to determining that there has been a previous interaction with the recommended content item, identifying an alternative recommended content item based on the first machine learning model output and the second machine learning model output and
modifying a display, corresponding to the control circuitry, to provide the alternative recommended content item.
2. The method of claim 1 , wherein analyzing the first feature vector and the second feature vector using the machine learning model comprises analyzing each of the first feature vector and the second feature vector using the machine learning model comprising at least one of a neural network or a Bayesian network.
3. The method of claim 1 , wherein transforming the at least one frame of the first content item and transforming the at least one frame of the second content item comprises transforming the at least one frame of the first content item and the at least one frame of the second content item using a Generalized Hough Transform.
4. The method of claim 1 , wherein the first feature vector is comprised of data corresponding to texture, shape intensity, and temporal data corresponding to the at least one frame of the first content item, and wherein the second feature vector is comprised of data corresponding to texture, shape intensity, and temporal data corresponding to the at least one frame of the second content item.
5. The method of claim 4 , wherein the temporal data corresponding to the first feature vector comprises data indicative of changes between respective frames of the at least one frame of the first content item, and wherein the temporal data corresponding to the second feature vector comprises data indicative of changes between respective frames of the at least one frame of the second content item.
6. The method of claim 1 , wherein generating the first feature vector comprises generating the first feature vector based on a mathematical operation of a plurality of feature vectors of respective frames of the at least one frame of the first content item, and wherein generating the second feature vector comprises generating the second feature vector based on a mathematical operation of a plurality of feature vectors of respective frames of the at least one frame of the second content item.
7. The method of 1 , further comprising:
receiving a first input indicative of the at least one frame of the first content item for which the recommended content item is to be provided; and
receiving a second input indicative of the at least one frame of the second content item for which the recommended content item is to be provided.
8. The method of claim 1 , wherein modifying the display further comprises:
identifying a plurality of frames of the recommended content item correlated to the at least one frame of the first video; and
displaying the plurality of frames simultaneously with an identifier of the recommended content item.
9. The method of claim 1 , wherein modifying the display further comprises:
identifying a plurality of frames of the recommended content item correlated to the at least one frame of the second video; and
displaying the plurality of frames simultaneously with an identifier of the recommended content item.
10. The method of claim 1 , wherein determining whether there has been a previous interaction with the recommended content item further comprises:
determining, based on the interaction data, whether the recommended content item was previously selected for playback.
11. A system comprising:
control circuitry configured to:
generate, using control circuitry, a first feature vector from at least one frame of a first content item, wherein generating the first feature vector comprises transforming the at least one frame of the first content item;
analyze, using the control circuitry, the first feature vector using a machine learning model to produce a first machine learning model output;
identify, based on the first machine learning model output, a second content item comprising at least one frame that is comprised of attributes corresponding to the first feature vector;
generate a second feature vector from at least one frame of the second content item, wherein generating the second feature vector comprises transforming the at least one frame of the second content item;
analyze, using the control circuitry the second feature vector using the machine learning model to produce a second machine learning model output; and
input/output circuitry configured to:
identify a recommended content item based on the first machine learning model output and the second machine learning model output;
determine whether the recommended content item has been generated as a recommendation previously;
determine, based on interaction data, whether there has been a previous interaction with the recommended content item;
in response to determining that there has been a previous interaction with the recommended content item, identify an alternative recommended content item based on the first machine learning model output and the second machine learning model output; and
modify a display, corresponding to the control circuitry, to provide the alternative recommended content item.
12. The system of claim 11 , wherein the control circuitry configured to analyze the first feature vector and the second feature vector using the machine learning model is further configured to analyze each of the first feature vector and the second feature vector using the machine learning model comprising at least one of a neural network or a Bayesian network.
13. The system of claim 11 , wherein the control circuitry configured to transform the at least one frame of the first content item and transform the at least one frame of the second content item is further configured to transform the at least one frame of the first content item and the at least one frame of the second content item using a Generalized Hough Transform.
14. The system of claim 11 , wherein control circuitry is further configured to the first feature vector such that it is comprised of data corresponding to texture, shape intensity, and temporal data corresponding to the at least one frame of the first content item, and generate the second feature vector such that it is comprised of data corresponding to texture, shape intensity, and temporal data corresponding to the at least one frame of the second content item.
15. The system of claim 14 , wherein the control circuitry is further configured to generate the first feature vector based on the temporal data corresponding to the first feature vector comprising data indicative of changes between respective frames of the at least one frame of the first content item, and wherein the control circuitry configured to generate the second feature vector based on the temporal data corresponding to the second feature vector comprising data indicative of changes between respective frames of the at least one frame of the second content item.
16. The system of 11 , wherein the control circuitry configured to generate the first feature vector is further configured to generate the first feature vector based on a mathematical operation of a plurality of feature vectors of respective frames of the at least one frame of the first content item, and wherein the control circuitry configured to generate the second feature vector is further configured to generate the second feature vector based on a mathematical operation of a plurality of feature vectors of respective frames of the at least one frame of the second content item.
17. The system of claim 11 , wherein the control circuitry is further configured to:
receive a first input indicative of the at least one frame of the first content item for which the recommended content item is to be provided; and
receive a second input indicative of the at least one frame of the second content item for which the recommended content item is to be provided.
18. The system of claim 11 , wherein the control circuitry configured to modify the display is further configured to:
identify a plurality of frames of the recommended content item correlated to the at least one frame of the first video; and
display the plurality of frames simultaneously with an identifier of the recommended content item.
19. The system of claim 11 , wherein the control circuitry configured to modify the display is further configured to:
identify a plurality of frames of the recommended content item correlated to the at least one frame of the second video; and
display the plurality of frames simultaneously with an identifier of the recommended content item.
20. The system of claim 11 , wherein the control circuitry configured to determine whether there has been a previous interaction with the recommended content item is further configured to:
determine, based on the interaction data, whether the recommended content item was previously selected for playback.