Indoor human activity dataset for millimeter-wave radar

This dataset consists of raw millimeter-wave radar data collected in indoor scenarios, with a unified file format of .bin. It is designed to support research on single-person and multi-person action recognition. Data collection covers two typical indoor scenarios, with dimensions of approximately 6m\u00d78m and 2.5m\u00d74m respectively. The scenes feature static backgrounds composed of walls and a small amount of furniture, eliminating dynamic interference factors.The hardware platform for data collection comprises Texas Instruments (TI) AWR2243 radar sensor and DCA1000EVM data acquisition card. The radar sensor is fixedly mounted on a table approximately 0.8m above the ground, ensuring a stable acquisition perspective. A total of 9 volunteers participated in data collection, with ages ranging from 20 to 30 years old and heights from 160cm to 185cm, covering the characteristic features of the common young adult population.The collected actions include walking, sitting, falling, squatting-standing, standing, and multiple gesture actions in single-person scenarios; as well as combined actions such as two people walking, one person walking and one sitting, one person sitting and one falling in multi-person scenarios. All actions are performed within the effective detection range of 1\u20138 meters in front of the radar. Each action segment lasts approximately 2\u20136 seconds, and the radar parameters are clearly configured (e.g., start frequency of 77GHz, chirp bandwidth of 3942.72MHz), ensuring the consistency and repeatability of the data.