Dataset · tanganke
SUN397 scene recognition benchmark (108,754 photos)
tanganke/sun397
Classic dataset of over 108k images for classifying 397 scene categories.
What's inside
A subset of the SUN database for scene recognition featuring 108,754 JPG images distributed across 397 English categories, with at least 108 examples per class. Category names are listed in ClassName.txt. Intended strictly for research purposes.
What you can do with it
- Training and evaluating scene and environment classification models
- Benchmarking visual encoders and zero-shot classifiers (e.g., CLIP)
- Fine-tuning lightweight models and LoRA for photo archive organization
- Tasks
- Image classification
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 33.4K
- Likes
- 4
Models trained on it
- tanganke/clip-vit-base-patch32_sun397Embeddings 3.6K
- tanganke/clip-vit-large-patch14_sun397Embeddings 765
- tanganke/clip-vit-base-patch32_TA8_mtl 101
- tanganke/clip-vit-large-patch14_TA8_mtl 22
- BlakePeavy/photo-identifier-v3Image classification 14
- AmeenAhmed2/zicoZero shot classification 0
- theblackcat102/clip-vit-base-patch16-task-experts-ep10Image classification 0
- theblackcat102/clip-vit-large-patch14-task-experts-ep10Image classification 0
How to load it
from datasets import load_dataset
ds = load_dataset("tanganke/sun397")