This dataset as been annotated by myself in the Pascal-VOC format using LabelImg.
The dataset is composed of 211 images of which 89.5% of them are for training and the rest of 10.5% are for validation. Since the dataset is relatively small, it is encouraged to fine-tune a preexisting model with this dataset.
| Train | Test | Total | |
|---|---|---|---|
| Size | 189 | 22 | 211 |
| Percentage | 89.5 % | 10.5 % | 100 % |
I tried to make this dataset as much mixed as I can by finding differents car angles.