Detecting and removing media modifications for identification services and copyright compliance

A computer readable medium comprising instructions that, when executed by a processing device, cause the processing device to perform operations comprising:

• processing, using a trained machine learning model provided by a computing device, a representation of a media content item to determine whether the media content item is a modified media content item, wherein the trained machine learning model outputs an indication of one or more modifications of the media content item;
• determining whether the one or more modifications satisfy a criterion;
• responsive to determining that the one or more modifications satisfy the criterion, performing a first operation of one or more operations on the media content item to reverse a first modification of the one or more modifications to generate an updated media content item, wherein the first modification has a highest likelihood of being present in the media content item;
• comparing the updated media content item to known media content items to identify known material contained therein;
• responsive to determining, based on the comparison, that the media content item does not contain known material, identifying, by the processing device, based on the indication, a second modification of the one or more modifications that has a second highest likelihood of being present in the media content item;
• performing, by the processing device, on the media content item, a second operation to reverse the second modification to generate a second updated media content item;
• and
• comparing, by the processing device, the second updated media content item to the known media content items to identify known material contained therein.

A system comprising:

• a memory;
• and
• a processing device operatively coupled to the memory, the processing device to:
• process a representation of a media content item using a trained machine learning model to determine whether the media content item is a modified media content item, wherein the trained machine learning model outputs an indication of one or more modifications of the media content item;
• determine whether the one or more modifications satisfy a criterion;
• responsive to determining that the one or more modifications satisfy the criterion, perform a first operation of one or more operations on the media content item to reverse a first modification of the one or more modifications to generate an updated media content item, wherein the first modification has a highest likelihood of being present in the media content item;
• compare the updated media content item to known media content items to identify known material contained therein;
• responsive to determining, based on the comparison, that the media content item does not contain known material, identify, by the processing device, based on the indication, a second modification of the one or more modifications that has a second highest likelihood of being present in the media content item;
• perform, by the processing device, on the media content item, a second operation to reverse the second modification to generate a second updated media content item;
• and
• compare, by the processing device, the second updated media content item to the known media content items to identify known material contained therein.

A method comprising:

• processing a representation of a media content item using a trained machine learning model to determine whether the media content item is a modified media content item, wherein the trained machine learning model outputs an indication of one or more modifications of the media content item;
• determining whether the one or more modifications satisfy a criterion;
• responsive to determining that the one or more modifications satisfy the criterion, performing one or more operations on the media content item in view of the one or more modifications to generate an updated media content item;
• and
• comparing the updated media content item to known media content items to identify known material contained therein by:
• comparing a digital fingerprint of the updated media content item to a plurality of additional digital fingerprints of the known media content items;
• identifying a match between the digital fingerprint and an additional digital fingerprint of the additional digital fingerprints, wherein the additional digital fingerprint is for a segment of a known media content item of the known media content items;
• and
• determining that the media content item comprises an instance of the known media content item.