Mastering Computer Vision with TensorFlow 2.x

PoseNet implementation

So far, we've discussed the theory behind PoseNet's top-down and bottom-up approaches. In this section, we will use PoseNet to identify these actions. Details of how to implement the PoseNet model can be found at https://github.com/tensorflow/tfjs-models/tree/master/posenet. Take a look at this link to understand the documentation of PoseNet.

Next, we will perform a live demo. This live demo, which is completed using a webcam, can be started by typing the following link in your web browser: https://storage.googleapis.com/tfjs-models/demos/posenet/camera.html.

Although the top-down and bottom-up approaches use the ResNet-101 model, the PoseNet model uses MobileNetV1 or ResNet-50. The difference between them is illustrated ...