Overhead ToF Depth and Wrist-Worn IMU Dataset for Identity-Verified Gesture Recognition in Smart-Room Environments

This dataset contains synchronized Time-of-Flight (ToF) depth camera recordings and wrist-worn IMU (MPU-6050) measurements collected for research on privacy-preserving, identity-verified gesture recognition in smart-room environments. Data were captured using a ceiling-mounted ESPROS TOFcam660 depth camera (320×240 pixels, overhead configuration at approximately 1.9 m height) and a wrist-worn 6-axis IMU transmitting at 50 Hz via WiFi UDP. The dataset comprises two components: Gesture templates: 493 individual gesture recordings (180 circle, 193 arm_up_down, 120 out-of-distribution) stored as normalized NumPy arrays (.npy) and their respective ROS bag files (.bag) indluding synchronized depth image streams and raw IMU measurements, recorded by 11 participants (5F/6M, heights 1.58–1.93 m). Templates can be found under "gesture_templates/" and their rosbags under "gesture_rosbags/". System evaluation recordings: 57 ROS bag files (.bag) capturing full interaction sessions (28 single-agent, 29 multi-agent) totaling approximately 2408 seconds, including synchronized depth image streams and raw IMU measurements. System rosbags can be found under folder "system_rosbags/" All recordings were made in an indoor environment under varying lighting conditions (natural, artificial, and combined). The camera field of view covers approximately 5.23 m × 3.02 m. No personally identifiable information is stored; depth maps contain no color, texture, or facial features, and participant identities appear only as anonymous integer tracker IDs that reset between sessions. To the best of our knowledge, this is the first publicly available dataset for gesture recognition from an overhead ToF depth sensor configuration with synchronized IMU data. It is intended to support future research on privacy-preserving gesture recognition, overhead depth-based human tracking, and multi-modal sensor fusion for identity verification.