Mastering Computer Vision with TensorFlow 2.x

Bottom-up approach

The second paper is titled PersonLab: Person Pose Estimation and Instance Segmentation with a Bottom-Up, Part-Based, Geometric Embedding Model, and was written by many of the same authors of the first paper; namely, George Papandreou, Tyler Zhu, Liang-Chieh Chen, Spyros Gidaris, Jonathan Tompson, and Kevin Murphy. You can find the paper at https://arxiv.org/abs/1803.08225.

In this box-free, bottom-up approach, the authors use a convolutional neural network to detect individual key points and their relative displacements to group key points into person pose instances. In addition, a geometric embedding descriptor was designed to determine person segmentations. The model is trained using the ResNet-101 and ResNet-152 architectures. ...