Gym Gesture Classification Using IMU Sensor Dataset

This dataset contains raw Inertial Measurement Unit (IMU) recordings for human activity recognition in strength training exercises, collected using a custom wearable device based on the Arduino Nano 33 BLE. The device was worn on the wrist and equipped with a 6-axis IMU (accelerometer and gyroscope), sampled at 100 Hz. Data was collected from five exercises commonly used in fitness training: chest press, chest fly, lat pulldown, tricep extension, and seated row.The cohort includes four trained athletes with more than six months of consistent exercise experience and one novice athlete with less than six months of experience, enabling analysis of cross-user generalization. Each participant performed three sets of ten repetitions per exercise, resulting in a total of 750 recorded movements.The dataset is stored in CSV format with the following headers: (athlete_id, exercise_type, weight_kg, set_number, rep_number, timestamp, ax, ay, az, gx, gy, gz), where accelerometer (ax, ay, az) and gyroscope (gx, gy, gz) signals capture movement dynamics. This dataset is suitable for research in TinyML, wearable computing, human activity recognition, and data augmentation strategies for cross-user performance generalization.