Footwear is typically worn during a range of daily activities, offering a valuable opportunity for integrating technologies like human pose estimation using embedded sensors. This study introduces a novel method of 3D human pose estimation using foot pressure data captured by a low-cost, high-resolution smart insole equipped with over 600 pressure sensors per foot. In contrast to previous works that used carpet-type sensors, which are limited to functioning only within a localized environment, our wireless smart insole enables pose estimation regardless of the user's location.We collect synchronized tactile and visual data across various actions. Utilizing a camera-based pose estimation model for supervision, we design a deep neural network to predict 3D human poses using only foot pressure data. Furthermore, integrating a simple linear classifier with our model’s learned representations achieves successful classification of various daily activities.