Overview

The media data processing APIs are described below. The APIs of different versions cannot be used together.

• Media data processing operator APIs: These APIs are suitable for AI frameworks. Their input and output data is based on tensor parameters, and asynchronous execution on streams is supported. In addition, these APIs can be used together with Neural Network (NN) operators on the same stream for collaborative processing.
• Media data processing V2 APIs: These APIs are suitable for computer vision applications. Their input and output data is based on image parameters. Multiple image formats such as YUV420 and RGB are supported, providing richer and more flexible parameter configuration capabilities.
• Media data processing V1 APIs: These are media APIs of an earlier version. They are no longer evolved and are replaced by the preceding two sets of APIs.