Dataset with the paper: FlexSense: Flexible Infrastructure Sensors for Traffic Perception. Please refer to the paper for more information. If you find this dataset or the paper relevant for your work, please consider citing our paper:
@INPROCEEDINGS{10422616,
author={Han, Longfei and Kefferpütz, Klaus and Elger, Gordon and Beyerer, Jürgen},
booktitle={2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC)},
title={FlexSense: Flexible Infrastructure Sensors for Traffic Perception},
year={2023},
volume={},
number={},
pages={3810-3816},
keywords={Point cloud compression;Three-dimensional displays;Roads;Radar tracking;Robot sensing systems;Sensor systems;Software},
doi={10.1109/ITSC57777.2023.10422616}}
The dataset consists of 29 recorded sequences using an infrared camera and a 3D radar sensor. The sensors work at 20Hz. In total 9730 frames are recorded. 24998 3D bounding boxes are provided. Example:
Please download the data here: https://fordatis.fraunhofer.de/handle/fordatis/344
├── IVI_Dataset_V1.0_FlexSense
├── calib
├── image
├── label
├── tracking
└── radar
calib contains the intrinsic calibration results of the IR camera and the extrinsic transformation between the camera and the radar.
Images are stored in the image folder.
label contains the label for 3D object detection.
tracking contains the label for 3D object tracking.
The bin files in folder radar contain the point cloud from the Continental ARS548 imaging radar.
The labeling is done using the 3D-BAT tool. In the labeling process, we tried our best to place the bounding box to match object extend in both 2D image plane and 3D point cloud world. If some of the 3D points don't perfectly fall into the bounding box, due to the sparsity of the point cloud, we have to consider them to be outlier.
Training set 6937 frames, Validation set 2793 frames.
| Sequences | Length (Frames) | Train/Val | Accumulated |
|---|---|---|---|
| Intersection sequence 1 | 374 | Train | |
| Intersection sequence 2 | 338 | Train | |
| Intersection sequence 3 | 374 | Train | |
| Intersection sequence 4 | 393 | Train | |
| Intersection sequence 5 | 392 | Train | |
| Intersection sequence 6 | 319 | Train | |
| Intersection sequence 7 | 393 | Train | |
| Intersection sequence 8 | 391 | Train | |
| Intersection sequence 9 | 391 | Train | |
| Intersection sequence 10 | 346 | Train | 3711 |
| Road sequence 1 | 292 | Train | |
| Road sequence 2 | 366 | Train | |
| Road sequence 3 | 237 | Train | |
| Road sequence 4 | 242 | Train | |
| Road sequence 5 | 289 | Train | |
| Road sequence 6 | 305 | Train | |
| Road sequence 7 | 353 | Train | |
| Road sequence 8 | 228 | Train | |
| Road sequence 9 | 238 | Train | |
| Road sequence 10 | 320 | Train | |
| Road sequence 11 | 356 | Train | 6937 |
| Intersection sequence 11 | 375 | Val | |
| Intersection sequence 12 | 384 | Val | |
| Intersection sequence 13 | 388 | Val | |
| Intersection sequence 14 | 385 | Val | 8469 |
| Road sequence 12 | 342 | Val | |
| Road sequence 13 | 284 | Val | |
| Road sequence 14 | 371 | Val | |
| Road sequence 15 | 264 | Val | 9730 |
| Values | Name | Description |
|---|---|---|
| 1 | type | Describes the type of object: 'Car', 'Van', 'Truck','Pedestrian', 'Person_sitting', 'Cyclist', 'Tram','Misc' or 'DontCare' |
| 1 | truncated | Float from 0 (non-truncated) to 1 (truncated), where truncated refers to the object leaving image boundaries |
| 1 | occluded | Integer (0,1,2,3) indicating occlusion state: 0 = fully visible, 1 = partly occluded, 2 = largely occluded, 3 = unknown |
| 1 | alpha | Observation angle of object, ranging [-pi..pi] |
| 4 | bbox | 2D bounding box of object in the image (0-based index):contains left, top, right, bottom pixel coordinates |
| 3 | dimensions | 3D object dimensions: height, width, length (in meters) |
| 3 | location | 3D object location x,y,z in camera coordinates (in meters) |
| 1 | rotation_y | Rotation ry around Y-axis in camera coordinates [-pi..pi] |
| (1) | score | Only for results: Float, indicating confidence in detection, needed for p/r curves, higher is better. |
| Values | Name | Description |
|---|---|---|
| 1 | frame | Frame within the sequence where the object appearers |
| 1 | track id | Unique tracking id of this object within this sequence |
| 1 | type | Describes the type of object: 'Car', 'Van', 'Truck', 'Pedestrian', 'Person_sitting', 'Cyclist', 'Tram','Misc' or 'DontCare' |
| 1 | truncated | Float from 0 (non-truncated) to 1 (truncated), where truncated refers to the object leaving image boundaries. Truncation 2 indicates an ignored object (in particular in the beginning or end of a track) introduced by manual labeling. |
| 1 | occluded | Integer (0,1,2,3) indicating occlusion state: 0 = fully visible, 1 = partly occluded 2 = largely occluded, 3 = unknown |
| 1 | alpha | Observation angle of object, ranging [-pi..pi] |
| 4 | bbox | 2D bounding box of object in the image (0-based index):contains left, top, right, bottom pixel coordinates |
| 3 | dimensions | 3D object dimensions: height, width, length (in meters) |
| 3 | location | 3D object location x,y,z in camera coordinates (in meters) |
| 1 | rotation_y | Rotation ry around Y-axis in camera coordinates [-pi..pi] |
| (1) | score | Only for results: Float, indicating confidence in detection, needed for p/r curves, higher is better. |