Dataset · tanganke
Stanford Cars with Robustness Augmentations
tanganke/stanford_cars
Classic car image dataset augmented with distorted test versions to evaluate model reliability.
What's inside
The collection includes 8,144 training and 8,041 test car images with English labels (totaling between 10k and 100k records). This build features additional test splits of 8,041 images each with various distortions: high contrast, Gaussian and impulse noise, JPEG compression, and motion blur. File formats are not specified in the description.
What you can do with it
- Training fine-grained image classification models
- Testing computer vision robustness against noise, compression, and blur
- Tasks
- Image classification
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 24.5K
- Likes
- 34
Models trained on it
- tanganke/clip-vit-base-patch32_stanford-carsEmbeddings 3.7K
- tanganke/clip-vit-large-patch14_stanford-carsEmbeddings 892
- tanganke/clip-vit-base-patch32_TA8_mtl 101
- ash12321/sdxl-detector-resnet50Image classification 28
- tanganke/clip-vit-large-patch14_TA8_mtl 22
- kikogazda/TwinCar-196Image classification 0
- kikogazda/TwinCar-196-v2Image classification 0
- kikogazda/Efficient_NetV2_EditionImage classification 0
How to load it
from datasets import load_dataset
ds = load_dataset("tanganke/stanford_cars")