English | 简体中文
Boden Around View Monitor Dataset is an open source dataset for academic and industrial purpose, which consists of 20k desensitized images with fully labeled annotations.
The images in this dataset are around-view images synthesized from four fisheye cameras. The detailed description of images are as follows:
-
Resolution: 1024 x 1024 pixels
-
View Range: 15m x 15m
-
Sample video of stitching images from 4 fisheye cameras
-
All images are anonymized to protect private information, such as number plates, faces and etc.
Sensor setup:
-
Fisheye cameras: SG2-IMX390C-GMSL2-H190s
- HFOV: 216°
- VFOV: 124°
- Resolution: 1920 x 1080 pixels
Scenes: The dataset was recorded in various indoor and outdoor parking sites. In order to cover scenes as much as we can, different illumination conditions and weather conditions were considered while collecting the data. Following are some examples:
- Underground Parking lot
- Indoor Parking Building
- Surface Parking lot
- Roadside Parking lot
Annotations: Annotations of the dataset have following forms:
- Parking Slot (released)
- Freespace (planning)
- Semantic Segmentation (planning)
- Detailed description of annotation specification can be seen here.
Planning
- release images and annotations
- release visualization script
- Download links for 2k sample images and annotations
- Baidu Cloud, code: 40pt
- Google Cloud
- 18K images and annotations:
- will be released soon.
- Application Only. Please send an application to our email address (company@bodenai.com). The template of application can be seen here.
The images and annotations of this dataset are licensed under a Creative Commons Attribution 4.0 License.
For mistakes in annotations submmision, professional support requests, or commercial cooperation, please contact us through the email address: company@bodenai.com.