├── datasetAgreement.pdf └── readme.md /datasetAgreement.pdf: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/ruixv/RadarEyes/HEAD/datasetAgreement.pdf -------------------------------------------------------------------------------- /readme.md: -------------------------------------------------------------------------------- 1 | # RadarEyes 2 | 3 | > mmWave (millimeter-wave) pointcloud dataset and preprocessing code 4 | 5 | 6 | 7 | - RadarEyes is a large-scale dataset focusing on indoor mmWave radar pointcloud. 8 | - [Paper Link: DREAM-PCD](https://arxiv.org/pdf/2309.15374.pdf) 9 | 10 | 11 | 12 | ## Dataset Introduction 13 | 14 | RadarEyes consists of aligned horizontally placed mmWave radar, vertically placed mmWave radar, LiDAR, and tracking cameras, capturing **120,000** frames of incoherent mmWave radar and **1,200,000** frames of coherent mmWave radar, along with corresponding poses and LiDAR pointcloud. 15 | 16 | ## Dataset Implementation 17 | 18 | ![RadarEyes](https://img-blog.csdnimg.cn/a568c502207a48eb89a9504ebe16b350.png) 19 | 20 | 21 | Our sensor platform, as shown above, is built on a remotely controllable vehicle and equipped with four sensors: 22 | 23 | **Two orthogonally placed mmWave radars:** Both are Texas Instruments AWR1843BOOST-EVM paired with a DCA1000-EVM for raw data capture. Each radar board has 3 transmitting antennas and 4 receiving antennas, operating at 77-81 GHz. Notably, RadarEyes contains raw ADC data and the source code for loading and processing the ADC data, which can significantly facilitate related research. 24 | 25 | **An 128-beam LiDAR:** LS-128 S2 1550nm [1], an automotive-grade hybrid solid-state LiDAR. 26 | 27 | **Camera and IMU:** ZED 2i Camera [2], an integrated device for simultaneous localization and mapping featuring optical odometry and IMU. The frame rate is set to 30fps, and the localization accuracy is under 1% closed-loop drift under intended use conditions. 28 | 29 | ### Hardware Configuration 30 | 31 | | Horizontal Radar | Vertical Radar | Lidar | 32 | |-----------------|---------------|------| 33 | | FMCW | FMCW | Pulse | 34 | | 10 | 100 | 10 | 35 | | $77.70-78.90GHz$ | $79.1-80.70GHz$ | $193THz$ | 36 | | 3 | 3 | - | 37 | | 4 | 4 | - | 38 | | 0.12 | 0.09375 | Lidar | 39 | | 15.99 | 12.0 | Lidar | 40 | | $11.3^\circ$ | $45^\circ$ | $0.09^\circ$ | 41 | | $\pm 50^\circ @6dB$ | $\pm 20^\circ @6dB$ | $\pm 60^\circ$ | 42 | | $45^\circ$ | $11.3^\circ$ | $0.1^\circ$ | 43 | | $\pm 20^\circ @6dB$ | $\pm 50^\circ @6dB$ | $\pm 12.5^\circ$ | 44 | | $0.12 m/s$ | $0.55 m/s$ | - | 45 | | $7.63 m/s$ | $4.47 m/s$ | - | 46 | 47 | ## How to access the dataset 48 | 49 | To obtain the dataset, please sign the [agreement](datasetAgreement.pdf), scan and send it to gengruixu@mail.ustc.edu.cn. You will receive a notification email which includes the download links of the dataset within two days. 50 | 51 | ### Preprocessing Code 52 | 53 | We provide Python scripts for data handling, including data loading, preprocessing, and visualization. These scripts are bundled with the part of the dataset provided via the download links in the confirmation email. 54 | 55 | ## Citation 56 | 57 | Please cite the following paper if this dataset is helpful to your research 58 | 59 | ```bibtex 60 | @article{geng2023dreampcd, 61 | title={DREAM-PCD: Deep Reconstruction and Enhancement of mmWave Radar Pointcloud}, 62 | author={Geng, Ruixu and Li, Yadong and Zhang, Dongheng and Wu, Jincheng and Gao, Yating and Hu, Yang and Chen, Yan}, 63 | journal={IEEE Transactions on Image Proceesing}, 64 | year={2023}, 65 | url={https://arxiv.org/abs/2309.15374} 66 | } 67 | ``` 68 | 69 | - Collection of works using RadarEyes for dataset 70 | - **DREAM-PCD** 71 | - **DiffRadar** 72 | 73 | --- 74 | 75 | **References:** 76 | 77 | 1. [Leishen Lidar](https://www.leishen-lidar.com/tof/158) 78 | 2. [ZED 2i Camera](https://www.stereolabs.com/zed-2i/) 79 | 80 | 81 | > Both the dataset and this README are currently being updated. Please stay tuned for updates on BaiduYun / OneDrive. 82 | 83 | --------------------------------------------------------------------------------