This is a small, hand-curated collection of 200–500 high-resolution color photos showing cars in two states: crashed and normal (undamaged). The images are split into 75% for training and approximately 25% for validation.
