fort-cyber/CFV-Dataset

CFV Dataset: Fine-Grained Predictions of Car Orientation from Images

Description

• Repository: CFV Dataset Repository
• Paper: CFV Dataset Paper

Dataset Summary

We present the CFV Dataset for estimating the car's orientation from images. Our dataset was obtained by recording cars while walking around them and annotating the frames with the pitch angle value in a semi-automatic manner. All images have the license plates anonymized.

Data Instance

One instance contains the following keys:

• image: The RGB image displaying the car
• identity: Unique integer for each recorded car
• angle: An integer ranging between 0-360, indicating the pitch angle value of the car in the image (in degrees)
• x1, y1, x2, y2: the bounding box of the car

Citation Information

If you utilize this dataset for any project or research, please cite our paper:

@article{catruna2023car,
  title={Car Full View Dataset: Fine-Grained Predictions of Car Orientation from Images},
  author={Catruna, Andy and Betiu, Pavel and Tertes, Emanuel and Ghita, Vladimir and Radoi, Emilian and Mocanu, Irina and Dascalu, Mihai},
  journal={Electronics},
  volume={12},
  number={24},
  pages={4947},
  year={2023},
  publisher={MDPI}
}