Go to
Multi-View Car Dataset
The dataset on this page contains 20 sequences of cars as they rotate by 360 degrees. There is one image approximately every 3-4 degrees. Using the time of capture information from the photos, it is possible to calculate the approximate rotation angle of the car.
The photos have been captured using a Nikon D70 camera on a tripod at the Geneva International Motor Show '08. The lens used is Nikkor 12-24mm DX f/4. The focal length was kept constant during a single sequence (showing a single car) but varies from sequence to sequence. The focus was set to manual, approx. at the hyperfocal distance.
All videos and images available from this page are copyrighted by CVLab - EPFL. You are free to use them for research purposes. If you use them to publish results, please cite the reference below.
Dataset
The thumbnails below show a set of frames from a sample sequence.
 |
 |
 |
 |
 |
 |
 |
 |
The preview video shows all the sequences from the dataset.
References
The dataset on this page has been used for our multiview object pose estimation algorithm described in the following paper:
M. Ozuysal, V. Lepetit and P.Fua, Pose Estimation for Category Specific Multiview Object Localization, Conference on Computer Vision and Pattern Recognition, Miami, FL, June 2009.
Secondary navigation
- Tree Structure Reconstruction
- Multi-view evaluation
- Stereo face database
- Multi-view stereo
- Multi-camera pedestrians video
- Multi-View Car Dataset
- Deformable Surface Reconstruction
- Local Keypoint and Optical Flow Evaluation
- IXMAS Actions - New Views and Occlusions
- Are Spatial and Global Constraints Really Necessary for Segmentation?
- Electron Microscopy Dataset
- Multi-view Multi-class Detection dataset
- Robust 3D Tracking with Descriptor Fields
- Polygonal Object Dataset
- RGB-D Pedestrian Dataset
- 3D Object Tracking from Monocular Images
- EPFL-RLC Multi-Camera Dataset