Methods and apparatus for identifying objects depicted in a video using extracted video frames in combination with a reverse image search engine
View Patent ↗Methods and apparatus are disclosed for identifying one or more objects (e.g., a logo, brand or commercial) depicted in a video. Textual information is identified from search results pages returned by a reverse image search engine for images extracted from the video. Base query records are generated corresponding to the search results pages that have textual information satisfying a base search term. Object query records are generated corresponding to the base query records that satisfy an object search term. A statistical criterion is applied to the object query records to identify an object depicted in the video. In some disclosed examples, the statistical criterion includes a threshold that is measured against the object query records and/or the base query records.
1. A method to identify an object depicted in a video, the method comprising:
capturing, by executing an instruction with at least one processor, screenshots of search results pages returned over a network from a reverse image search engine hosted by a remotely located server, the search results pages corresponding to respective ones of a plurality of image frames of the video that were processed by the reverse image search engine, the respective ones of the image frames being accessed by the reverse image search engine based on a uniform resource locator associated with the image frames, respective ones of the search results pages including dynamic data, the capturing of the screenshots of the search results pages converting the dynamic data into static data included in the screenshots;
identifying, by executing an instruction with the at least one processor, textual information presented in respective ones of the screenshots;
generating, by executing an instruction with the at least one processor, a plurality of base query records corresponding to respective ones of the screenshots that have respective textual information satisfying a base search term;
generating, by executing an instruction with the at least one processor, a plurality of object query records corresponding to respective ones of the base query records that satisfy an object search term;
determining, by executing an instruction with the at least one processor, whether a statistical criterion applied to the object query records is satisfied; and
in response to determining that the statistical criterion is satisfied, identifying, by executing an instruction with the at least one processor, the object depicted in the video based on the object search term.
2. The method as defined in claim 1 , wherein the identifying of the textual information presented in the respective ones of the screenshots includes executing an optical character recognition application.
3. The method as defined in claim 1 , wherein the statistical criterion includes a threshold based on a first number of the object query records including the object search term.
4. The method as defined in claim 1 , wherein the statistical criterion includes a threshold based on a first number of the object query records including the object search term within a first time period, the first time period based on a sampling rate at which the image frames were obtained from the video.
5. The method as defined in claim 1 , wherein the statistical criterion includes a threshold based on a first consecutive number of the object query records including the object search term.
6. The method as defined in claim 1 , wherein the statistical criterion includes a threshold based on a first percentage of a number of object query records including the object search term relative to a number of base query records including the base search term.
7. The method as defined in claim 1 , wherein the object depicted in the video includes at least one of a logo, a brand, or a commercial.
8. The method as defined in claim 1 , wherein the base search term is associated with at least one of a word or a phrase that is expected to be included in the respective ones of the search results pages returned by the reverse image search engine.
9. The method as defined in claim 1 , wherein the base search term is associated with at least one of a position or a threshold relative to a configuration of a layout that is expected to be included in the respective ones of the search results pages returned by the reverse image search engine.
10. The method as defined in claim 1 , wherein the respective ones of the base query records include the base search term and a portion of the textual information that is associated with the base search term in the corresponding screenshot, the portion either following the base search term in the corresponding screenshot, being on a same line as the base search term in the corresponding screenshot, or being part of a same string as the base search term in the corresponding screenshot.
11. The method as defined in claim 10 , further including:
performing, by executing an instruction with the at least one processor, a frequency analysis on the respective ones of the base query records, the frequency analysis including at least one of a word count analysis, a phrase count analysis, or a string comparison analysis; and
determining, by executing an instruction with the at least one processor, the object search term based on a result of the frequency analysis.
12. The method as defined in claim 11 , wherein the performing of the frequency analysis includes filtering out the base search term from the respective ones of the base query records.
13. A system to identify an object depicted in a video, the system comprising:
a screen capturer structured to capture screenshots of search results pages returned over a network from a reverse image search engine hosted by a remotely located server, the search results pages corresponding to respective ones of a plurality of image frames of the video that were processed by the reverse image search engine, the respective ones of the image frames being accessed by the reverse image search engine based on a uniform resource locator associated with the image frames, respective ones of the search results pages including dynamic data, the dynamic data being converted into static data included in the screenshots;
a text identifier structured to identify textual information presented in respective ones of the screenshots;
a base query record generator structured to generate a plurality of base query records corresponding to respective ones of the screenshots that have respective textual information satisfying a base search term;
an object query record generator structured to generate a plurality of object query records corresponding to respective ones of the base query records that satisfy an object search term; and
an object identifier structured to:
determine whether a statistical criterion applied to the object query records is satisfied; and
in response to determining that the statistical criterion is satisfied, identify the object depicted in the video based on the object search term, each of the screen capturer, the text identifier, the base query record identifier, the object query record identifier, and the object identifier being implemented via a logic circuit.
14. The system as defined in claim 13 , further including a video frame sampler structured to extract the image frames from the video.
15. The system as defined in claim 13 , wherein the text identifier is structured to execute an optical character recognition application to identify the textual information presented in the respective ones of the screenshots.
16. The system as defined in claim 13 , wherein the statistical criterion includes a threshold based on a first number of the object query records including the object search term.
17. The system as defined in claim 13 , wherein the statistical criterion includes a threshold based on a first number of the object query records including the object search term within a first time period, the first time period based on a sampling rate at which the image frames were obtained from the video.
18. The system as defined in claim 13 , wherein the statistical criterion includes a threshold based on a first consecutive number of the object query records including the object search term.
19. The system as defined in claim 13 , wherein the statistical criterion includes a threshold based on a first percentage of a number of object query records including the object search term relative to a number of base query records including the base search term.
20. The system as defined in claim 13 , wherein the object depicted in the video includes at least one of a logo, a brand, or a commercial.
21. A tangible computer readable storage medium comprising computer readable instructions that, when executed, cause a processor to at least:
capture screenshots of search results pages returned over a network from a reverse image search engine hosted by a remotely located server, the search results pages corresponding to respective ones of a plurality of image frames of a video that were processed by the reverse image search engine, the respective ones of the image frames being accessed by the reverse image search engine based on a uniform resource locator associated with the image frames, respective ones of the search results pages including dynamic data, the capturing of the screenshots of the search results pages converting the dynamic data into static data included in the screenshots;
identify textual information presented in respective ones of the screenshots;
generate a plurality of base query records corresponding to respective ones of the screenshots that have respective textual information satisfying a base search term;
generate a plurality of object query records corresponding to respective ones of the base query records that satisfy an object search term;
determine whether a statistical criterion applied to the object query records is satisfied; and
in response to determining that the statistical criterion is satisfied, identify an object depicted in the video based on the object search term.
22. The storage medium as defined in claim 21 , wherein the instructions further cause the processor to execute an optical character recognition application to identify the textual information presented in the respective ones of the screenshots.
23. The storage medium as defined in claim 21 , wherein the statistical criterion includes a threshold based on a first number of the object query records including the object search term.
24. The storage medium as defined in claim 21 , wherein the statistical criterion includes a threshold based on a first number of the object query records including the object search term within a first time period, the first time period based on a sampling rate at which the image frames were obtained from the video.
25. The storage medium as defined in claim 21 , wherein the statistical criterion includes a threshold based on a first consecutive number of the object query records including the object search term.
26. The storage medium as defined in claim 21 , wherein the statistical criterion includes a threshold based on a first percentage of a number of object query records including the object search term relative to a number of base query records including the base search term.
27. The storage medium as defined in claim 21 , wherein the object depicted in the video includes at least one of a logo, a brand, or a commercial.