Human-computer interaction method based on artificial intelligence

Abstract

The invention discloses a man-machine interaction method based on artificial intelligence. The man-machine interaction method comprises the following steps: establishing an identification model; inputting the video into a recognition model, and recognizing the video to obtain a dynamic gesture of a character in the video; the recognition model comprises a space channel sub-model and a time channel sub-model, the space channel sub-model processes space information for video frames, and the time channel sub-model processes time sequence information and motion feature information for video clips. The man-machine interaction method based on artificial intelligence has the advantages of being high in recognition precision, high in frame rate, high in speed and the like.

Description

technical field [0001] The invention relates to a human-computer interaction method based on artificial intelligence, in particular to a dynamic meeting gesture recognition method, which belongs to the technical field of image recognition and detection. Background technique [0002] In computer vision recognition, we can classify images and detect objects in images. At present, the model based on convolutional neural network has achieved the task of image cognition. However, this type of model has serious limitations in understanding image sequences, and cannot identify the semantic correlation between consecutive images, that is, it cannot respond to dynamic behaviors. to recognize or understand. [0003] But in the real world, most behaviors cannot be judged by static pictures. For example, a picture is extracted from the middle process of zooming out or zooming in. The static pictures are basically the same, even for humans, it is difficult to distinguish them. [0004]...

AI Technical Summary

Problems solved by technology

At present, the model based on convolutional neural network has achieved the task of image cognition. However, this type of model has serious limitations in understanding image sequences, and cannot identify the semantic correlation between consecutive images, that is, it cannot respond to dynamic behaviors. to recognize or understand

[0003] But in the real world, most behaviors cannot be judged by static pictures. For example, a picture is extracted from the middle process of zooming out or zooming in. The static pictures are basically the same, even for humans, it is difficult to distinguish

[0004] Although there are dynamic gesture recognition products or methods such as Kinect, they all need specific hardware equipment, so they do not have universality; in addition, such products or methods have higher requirements for users, and the operation before use complicated steps

[0005] Moreover, the traditional recognition method has low accuracy and stability for dynamic gesture recognition, slow recognition speed, and poor user experience

Images