Model-based data processing method and apparatus

A model-based data processing apparatus, comprising:

• at least one memory configured to store program code;
• and
• at least one processor configured to read the program code and operate as instructed by the program code, the program code comprising:
• sample obtaining code configured to cause the at least one processor to extract a first training sample set in order to obtain a second training sample set and extract, a query video, at least one video sample that is not included in the first raining sample set, the first training sample set comprising different types of video samples, the second training sample set obtained by combining the variety of types of video samples in the first training sample set;
• feature extraction code, configured to cause the at least one processor to process the second training sample set through an embedded layer network in an action recognition model to obtain a first frame feature sequence, the first frame feature sequence comprising a first plurality of frames;
• first processing code configured to cause the at least one processor to process the query video through the embedded layer network to obtain a second frame feature sequence, the second frame feature sequence comprising a second plurality of frames;
• time sequence code, configured to cause the at least one processor to process the first frame feature sequence through a time sequence relation network in the action recognition model to obtain a first time sequence relation descriptor;
• second processing code configured to cause the at least one processor process the second frame feature sequence through the time sequence relation network to obtain a second time sequence relation descriptor;
• and
• model training code configured to cause the at least one processor to adjust a model parameter of the action recognition model according to the first time sequence relation descriptor and the second time sequence relation descriptor, the adjusted action recognition model being configured to recognize an action in a video to be recognized.

A non-transitory computer-readable storage medium, storing a computer program that when executed by at least one processor causes the at least one processor to:

• extract a first training sample set in order to obtain a second training sample set and extract, as a query video, at least one video sample that is not included in the first training sample set, the first training sample set comprising a variety of types of video samples, the second training sample set obtained by combining the variety of types of video samples in the first training sample set;
• process the second training sample set through an embedded layer network in an action recognition model to obtain a first frame feature sequence, the first frame feature sequence comprising a first plurality of frames;
• process the query video through the embedded layer network to obtain a second frame feature sequence, the second frame feature sequence comprising a second plurality of frames;
• process the first frame feature sequence through a time sequence relation network in the action recognition model to obtain a first time sequence relation descriptor;
• process the second frame feature sequence through the time sequence relation network to obtain a second time sequence relation descriptor;
• and
• adjust a model parameter of the action recognition model according to the first time sequence relation descriptor and the second time sequence relation descriptor, the adjusted action recognition model being configured to recognize an action in a video to be recognized.