SUN397

archive 2025-07-28

The Scene UNderstanding (SUN) database contains 899 categories and 130,519 images. There are 397 well-sampled categories to evaluate numerous state-of-the-art algorithms for scene recognition.

Image Source: The Selection of Useful Visual Words in Class-Imbalanced Image Classification

Homepage Benchmarks Papers JSON

Benchmarks archive 2025-07-28

All 8 leaderboards whose dataset resolves to this page shown (sort by any header). "First row" is the archive's own first row at snapshot, in the archive's row order; nothing here re-ranks and metric direction is not asserted.

		First row (archive order)	Paper	Code
Prompt Engineering	SUN397	PromptKD Harmonic mean 82.60	PromptKD: Unsupervised Prompt Distillation for...	zhengli97/promptkd	14	Compare
Fine-Grained Image Classification	SUN397	µ2Net (ViT-L/16) Accuracy 84.8	An Evolutionary Approach to Dynamic Introduction of...	google-research/google-research	5	Compare
Scene Recognition	SUN397	FOSNet Accuracy 77.28	FOSNet: An End-to-End Trainable Deep Neural Network for...	—	2	Compare
Zero-Shot Learning	SUN397	ZLaP* Accuracy 71.4	Label Propagation for Zero-shot Classification with...	vladan-stojnic/zlap	2	Compare
Few-Shot Learning	SUN397	Variational Prompt Tuning Harmonic mean 78.51	Bayesian Prompt Learning for Image-Language Model Generalization	saic-fi/bayesian-prompt-learning	1	Compare
Image Classification	SUN397	TransBoost-ResNet50 Accuracy 95.94%	TransBoost: Improving the Best ImageNet Performance...	omerb01/transboost	1	Compare
Image Clustering	SUN397	TURTLE (CLIP + DINOv2) Accuracy 67.9	Let Go of Your Labels with Unsupervised Transfer	mlbio-epfl/turtle	1	Compare
Transductive Zero-Shot Classification	SUN397	ZLaP Accuracy 71.9	Label Propagation for Zero-shot Classification with...	vladan-stojnic/zlap	1	Compare

Papers archive 2025-07-28

25 shown of 25 papers with a leaderboard row on this dataset's benchmarks, newest first. The archive's own "papers using this dataset" list was never published, so this is the benchmark-backed subset; the archive's count for this dataset is 52. The Syntology column is from Syntology's graph (read 2026-09-24), stated per sample; it is not part of any archive number.

			Date	Samples run Syntology
MMRL++: Parameter-Efficient and Interaction-Aware Representation Learning for Vision-Language Models	1	1	15 May 2025	not harvested
MMRL: Multi-Modal Representation Learning for Vision-Language Models	1	1	11 Mar 2025	ran 0 of 1 samples (1 unverified)
HPT++: Hierarchically Prompting Vision-Language Models with Multi-Granularity Knowledge Generation and Improved Structure Modeling	2	1	27 Aug 2024	not harvested
Let Go of Your Labels with Unsupervised Transfer	1	1	11 Jun 2024	ran 3 of 4 samples (1 unverified; 4 pointer-only for licence)
Label Propagation for Zero-shot Classification with Vision-Language Models	1	3	5 Apr 2024	ran 1 of 2 samples (1 unverified)
Prompt Learning via Meta-Regularization	1	1	1 Apr 2024	ran 4 of 7 samples (3 unverified)
PromptKD: Unsupervised Prompt Distillation for Vision-Language Models	1	1	5 Mar 2024	not harvested
Learning Hierarchical Prompt with Structured Linguistic Knowledge for Vision-Language Models	2	1	11 Dec 2023	ran 3 of 7 samples (4 unverified)
DePT: Decoupled Prompt Tuning	1	1	14 Sep 2023	ran 4 of 8 samples (4 unverified; 8 pointer-only for licence)
Read-only Prompt Optimization for Vision-Language Few-shot Learning	1	1	29 Aug 2023	ran 4 of 6 samples (2 unverified)
Self-regulating Prompts: Foundational Model Adaptation without Forgetting	2	1	13 Jul 2023	ran 7 of 21 samples (14 unverified)
Consistency-guided Prompt Learning for Vision-Language Models	2	1	1 Jun 2023	ran 3 of 7 samples (4 unverified)
Learning Domain Invariant Prompt for Vision-Language Models	1	1	8 Dec 2022	not harvested
MaPLe: Multi-modal Prompt Learning	3	1	6 Oct 2022	ran 4 of 6 samples (2 unverified)
Bayesian Prompt Learning for Image-Language Model Generalization	1	1	5 Oct 2022	ran 4 of 6 samples (2 unverified)
TransBoost: Improving the Best ImageNet Performance using Deep Transduction	1	1	26 May 2022	not harvested
An Evolutionary Approach to Dynamic Introduction of Tasks in Large-scale Multitask Learning Systems	1	1	25 May 2022	not harvested
Bamboo: Building Mega-Scale Vision Dataset Continually with Human-Machine Synergy	2	1	15 Mar 2022	ran 3 of 3 samples (0 unverified; 3 pointer-only for licence)
Conditional Prompt Learning for Vision-Language Models	12	1	10 Mar 2022	ran 4 of 6 samples (2 unverified)
Vision Models Are More Robust And Fair When Pretrained On Uncurated Images Without Supervision	1	1	16 Feb 2022	not harvested
Self-Supervised Learning by Estimating Twin Class Distributions	2	1	14 Oct 2021	ran 5 of 15 samples (10 unverified)
With a Little Help from My Friends: Nearest-Neighbor Contrastive Learning of Visual Representations	4	1	29 Apr 2021	ran 4 of 5 samples (1 unverified)
Learning Transferable Visual Models From Natural Language Supervision	82	1	26 Feb 2021	ran 16 of 20 samples (4 unverified; 16 pointer-only for licence)
Semantic-Aware Scene Recognition	1	1	5 Sep 2019	ran 0 of 3 samples (3 unverified)
FOSNet: An End-to-End Trainable Deep Neural Network for Scene Recognition	0	1	17 Jul 2019	not harvested

Dataset loaders archive 2025-07-28

huggingface.co/datasets/1aurent/SUN397 huggingface/datasetstfpytorchjax

pytorch.org/vision/stable/generated/torchvision.datasets.SUN397.html pytorch/visionpytorch

www.tensorflow.org/datasets/catalog/sun397 tensorflow/datasetstfjax

3 loaders as listed in the archive; links are outbound and not re-checked here.

Tasks archive 2025-07-28

Image Classification Zero-Shot Learning Few-Shot Learning Image Clustering Fine-Grained Image Classification Transductive Zero-Shot Classification Prompt Engineering Few-Shot Learning - 4 shots Scene Recognition

License archive 2025-07-28

No licence recorded in the archive. Absence here is not a statement about the dataset's terms.

Modalities archive 2025-07-28

Images

Languages archive 2025-07-28

No language tagged.

Variants archive 2025-07-28

SUN397

1 variant name, as the archive lists them.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections