No abstract

At-home exercise monitoring is vital to applications like rehabilitative care and physical therapy. In this work, we use millimeter-wave signal reflections to assess the exercise, where we classify the exercise type by designing a supervised deep learning model, and estimate the number of repetitions by leveraging phase information embedded in the reflections. CCS CONCEPTS• Human-centered computing → Ubiquitous and mobile computing; • Computing methodologies → Neural networks.

No abstract

3D Point Cloud (PCD) is an efficient machine representation for surrounding environments and has been used in many applications. But a fast reconstruction of complete PCD for large environments remains a challenge. We propose AutoPCD, a machine-learning model that reconstructs complete PCDs, under sensor occlusion and poor lighting conditions. AutoPCD splits the PCD into multiple parts, approximates them by several 3D planes, and independently learns the plane features for reconstruction. We have experimentally evaluated AutoPCD in a large indoor hallway environment. CCS CONCEPTS• Human-centered computing → Ubiquitous and mobile computing systems and tools; • Computing methodologies → Machine learning approaches.