EgoPAT3Dv2: 3D Action Target Prediction from Egocentric Vision for Human-Robot Interaction

Predicting 3D action targets from egocentric vision enhances human-robot interaction safety and efficiency.