Motion Posture Detection Based on Image Recognition through Deep Learning Algorithm

In computer vision, image recognition and action detection are two very important tasks. These two tasks can be viewed as two different techniques: feature-based methods and model-based methods. Image recognition techniques represent objects or objects in an image as a vector of numbers, which can be viewed as a set of features. However, when there is a certain relationship between these vectors and poses or actions, these vectors are called feature vectors. The article adopts a method based on deep learning algorithm to realize motion posture detection. This method can collect video stream through camera for motion detection. The experimental results show that the detection accuracy of the motion posture detection method based on image recognition can reach up to 94.5% through the deep learning algorithm.