Volume 13 | Issue 4
Volume 13 | Issue 4
Volume 13 | Issue 4
Volume 13 | Issue 4
Volume 13 | Issue 4
This study presents a novel approach for generating joint angular displacement maps (JADMs) by computing the distance and angle between pairs of joints. The resulting JAD matrix is color-coded to represent the JADMs. The JADMs are then inputted into a lightweight deep network, which achieves an average recognition accuracy of 84% over joint distance maps (JADs).