VARISHTA-MM50: A Synchronized Sensor-Video Dataset for Activity Monitoring and Fall Detection in Indian Elderly Populations

VARISHTA-MM50 is a publicly available synchronized dual-smartphone multimodal dataset for human activity recognition and fall detection in Indian elderly populations. 50 participants (aged 60-90 years, 18 female, 32 male) across five data collection batches (July 2025 - June 2026). 9 activity classes: 6 ADL (Walking, Standing, Sitting, Jogging, Stair_Climbing, Stair_Descending) and 3 falls (Trip_n_Fall, Sit_n_Fall, Walk_n_Fall). Each PersonN folder contains: Accelerometer, Gyroscope, Orientation, sensor_merged CSVs and synchronized video MP4. Sensor data at 50Hz, video at 1080p/30fps. Collected using low-cost dual-smartphone protocol. Ethics approved by IEC, SPIT Mumbai.