Brand safety in video content

A system comprising:

• a processing device;
• and
• a non-transitory computer-readable medium communicatively coupled to the processing device, wherein the processing device is configured to execute program code stored in the non-transitory computer-readable medium and thereby perform operations comprising:
• obtaining a video, the video including a plurality of scenes, each scene including a plurality of video frames;
• extracting, using a first set of models implemented by the one or more processing devices, frame-level features from each of two or more video frames in each scene of the plurality of scenes;
• generating, using a second set of models implemented by the one or more processing devices, scene-level features for each scene of the plurality of scenes based on the frame-level features extracted from the two or more video frames in each scene of the plurality of scenes;
• generating, using a third set of models implemented by the one or more processing devices, video-level features in the video based on the scene-level features generated for each scene of the plurality of scenes;
• and
• determining a brand safety score for the video based on the video-level features, the brand safety score indicating whether it is safe to associate a brand with the video.

A system comprising:

• means for obtaining a video, the video including a plurality of scenes, each scene including a plurality of video frames;
• means for extracting, using a first set of models implemented by one or more processing devices, frame-level features from each of two or more video frames in each scene of the plurality of scenes;
• means for generating, using a second set of models implemented by the one or more processing devices, scene-level features for each scene of the plurality of scenes based on the frame-level features extracted from the two or more video frames in each scene of the plurality of scenes;
• means for generating, using a third set of models implemented by the one or more processing devices, video-level features in the video based on the scene-level features generated for each scene of the plurality of scenes;
• and
• means for determining a brand safety score for the video based on the video-level features, the brand safety score indicating whether it is safe to associate a brand with the video.