Datasets › iNaturalist

iNaturalist

Introduced by Grant Van Horn et al. in The iNaturalist Species Classification and Detection Dataset1 Jan 2018 archive 2025-07-28

The iNaturalist 2017 dataset (iNat) contains 675,170 training and validation images from 5,089 natural fine-grained categories. Those categories belong to 13 super-categories including Plantae (Plant), Insecta (Insect), Aves (Bird), Mammalia (Mammal), and so on. The iNat dataset is highly imbalanced with dramatically different number of images per category. For example, the largest super-category “Plantae (Plant)” has 196,613 images from 2,101 categories; whereas the smallest super-category “Protozoa” only has 381 images from 4 categories.

Source: Large Scale Fine-Grained Categorization and Domain-Specific Transfer Learning Image Source: https://github.com/visipedia/inat_comp/tree/master/2017

Homepage Benchmarks Papers JSON

Benchmarks archive 2025-07-28

All 12 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
Image Classification	iNaturalist 2018	OmniVec2 Top-1 Accuracy 94.6	OmniVec2 - A Novel Transformer based Network for Large...	—	60	Compare
Long-tail Learning	iNaturalist 2018	LIFT (ViT-L/14@336px) Top-1 Accuracy 87.4%	Long-Tail Learning with Foundation Model: Heavy Fine-Tuning Hurts	shijxcs/lift	43	Compare
Image Classification	iNaturalist 2019	Hiera-H (448px) Top-1 Accuracy 88.5	Hiera: A Hierarchical Vision Transformer without the...	huggingface/pytorch-image-models +3	22	Compare
Image Classification	iNaturalist	AIMv2-3B (448 res) Top 1 Accuracy 85.9	Multimodal Autoregressive Pre-training of Large Vision Encoders	apple/ml-aim	19	Compare
Image Retrieval	iNaturalist	Unicom+ViT-L@336px R@1 88.9	Unicom: Universal and Compact Representation Learning...	OML-Team/open-metric-learning +2	10	Compare
Image Generation	iNaturalist 2019	StyeGAN2 + NoisyTwins FID 11.46	NoisyTwins: Class-Consistent and Diverse Image...	val-iisc/NoisyTwins	2	Compare
Few-Shot Image Classification	iNaturalist (227-way multi-shot)	LaplacianShot Accuracy 74.97	Laplacian Regularized Few-Shot Learning	sicara/easy-few-shot-learning +1	1	Compare
Few-Shot Image Classification	iNaturalist 2018 - 1-shot	MAWS (ViT-2B) Top 1 Accuracy 35.5	The effectiveness of MAE pre-pretraining for...	facebookresearch/maws	1	Compare
Few-Shot Image Classification	iNaturalist 2018 - 5-shot	MAWS (ViT-2B) Top 1 Accuracy 72.8	The effectiveness of MAE pre-pretraining for...	facebookresearch/maws	1	Compare
Few-Shot Image Classification	iNaturalist 2018 - 10-shot	MAWS (ViT-2B) Top 1 Accuracy 80.3	The effectiveness of MAE pre-pretraining for...	facebookresearch/maws	1	Compare
Fine-Grained Image Classification	iNaturalist	TASN Top 1 Accuracy 68.2	Looking for the Devil in the Details: Learning Trilinear...	researchmm/tasn	1	Compare
Image Classification	iNat2021-mini	WaveMix-256/16 (level 2) Top 1 Accuracy 61.75	WaveMix: A Resource-efficient Neural Network for Image Analysis	pranavphoenix/WaveMix	1	Compare

Papers archive 2025-07-28

30 shown of 84 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 603. 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
Multimodal Autoregressive Pre-training of Large Vision Encoders	1	5	21 Nov 2024	ran 0 of 3 samples (3 unverified; 3 pointer-only for licence)
Adaptive Parametric Activation	1	1	11 Jul 2024	not harvested
Harnessing Hierarchical Label Distribution Variations in Test Agnostic Long-tail Recognition	1	1	13 May 2024	ran 1 of 1 samples (0 unverified; 1 pointer-only for licence)
DeiT-LT Distillation Strikes Back for Vision Transformer Training on Long-Tailed Datasets	2	2	3 Apr 2024	not harvested
DenseNets Reloaded: Paradigm Shift Beyond ResNets and ViTs	3	8	28 Mar 2024	ran 3 of 3 samples (0 unverified)
Probabilistic Contrastive Learning for Long-Tailed Visual Recognition	1	1	11 Mar 2024	ran 6 of 9 samples (3 unverified)
Learning Semantic Proxies from Visual Prompts for Parameter-Efficient Fine-Tuning in Deep Metric Learning	1	1	4 Feb 2024	ran 3 of 5 samples (2 unverified)
OmniVec2 - A Novel Transformer based Network for Large Scale Multimodal and Multitask Learning	0	1	1 Jan 2024	not harvested
OmniVec: Learning robust representations with cross modal sharing	0	1	7 Nov 2023	not harvested
Long-Tail Learning with Foundation Model: Heavy Fine-Tuning Hurts	1	3	18 Sep 2023	ran 5 of 8 samples (3 unverified; 8 pointer-only for licence)
MDCS: More Diverse Experts with Consistency Self-distillation for Long-tailed Recognition	1	1	19 Aug 2023	ran 7 of 8 samples (1 unverified; 8 pointer-only for licence)
Hiera: A Hierarchical Vision Transformer without the Bells-and-Whistles	4	3	1 Jun 2023	ran 0 of 6 samples (6 unverified)
Long-Tailed Recognition by Mutual Information Maximization between Latent Features and Ground-Truth Labels	1	2	2 May 2023	ran 1 of 1 samples (0 unverified)
Unicom: Universal and Compact Representation Learning for Image Retrieval	3	1	12 Apr 2023	ran 3 of 6 samples (3 unverified; 6 pointer-only for licence)
NoisyTwins: Class-Consistent and Diverse Image Generation through StyleGANs	1	1	12 Apr 2023	not harvested
The effectiveness of MAE pre-pretraining for billion-scale pretraining	1	4	23 Mar 2023	not harvested
Escaping Saddle Points for Effective Generalization on Class-Imbalanced Data	1	1	28 Dec 2022	not harvested
InternImage: Exploring Large-Scale Vision Foundation Models with Deformable Convolutions	3	1	10 Nov 2022	ran 2 of 4 samples (2 unverified)
Generalized Parametric Contrastive Learning	4	5	26 Sep 2022	not harvested
A Continual Development Methodology for Large-scale Multitask Dynamic ML Systems	1	1	15 Sep 2022	not harvested
Improving GANs for Long-Tailed Data through Group Spectral Regularization	1	1	21 Aug 2022	ran 0 of 3 samples (3 unverified)
Conviformers: Convolutionally guided Vision Transformer	1	1	17 Aug 2022	not harvested
Balanced Contrastive Learning for Long-Tailed Visual Recognition	1	1	19 Jul 2022	ran 1 of 1 samples (0 unverified)
ViT-NeT: Interpretable Vision Transformers with Neural Tree Decoder	1	1	17 Jul 2022	not harvested
Hierarchical Average Precision Training for Pertinent Image Retrieval	2	2	5 Jul 2022	ran 6 of 9 samples (3 unverified)
WaveMix: A Resource-efficient Neural Network for Image Analysis	1	1	28 May 2022	not harvested
On the Eigenvalues of Global Covariance Pooling for Fine-grained Visual Recognition	1	1	26 May 2022	ran 3 of 17 samples (14 unverified)
MixMAE: Mixed and Masked Autoencoder for Efficient Pretraining of Hierarchical Vision Transformers	1	3	26 May 2022	not harvested
Nested Collaborative Learning for Long-Tailed Visual Recognition	1	1	29 Mar 2022	ran 2 of 2 samples (0 unverified)
Long-Tailed Recognition via Weight Balancing	2	1	27 Mar 2022	ran 5 of 14 samples (9 unverified; 13 pointer-only for licence)

The full list of 84 is in the JSON twin.

Dataset loaders archive 2025-07-28

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

www.tensorflow.org/datasets/catalog/i_naturalist2017 tensorflow/datasetstfjax

visipedia/inat_comp

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

Tasks archive 2025-07-28

Image Classification Image Generation Few-Shot Image Classification Image Retrieval Fine-Grained Image Classification Long-tail Learning Test Agnostic Long-Tailed Learning

License archive 2025-07-28

Custom (non-commercial)

Modalities archive 2025-07-28

Images

Languages archive 2025-07-28

No language tagged.

Variants archive 2025-07-28

iNat2021-mini
iNat2021
iNaturalist Fine-Grained Geolocation
iNaturalist 2018 - 5-shot
iNaturalist 2018 - 1-shot
iNaturalist 2018 - 10-shot
iNaturalist 2019
iNaturalist 2018
iNaturalist (227-way multi-shot)
iNaturalist

10 variant names, 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