Browse State-of-the-Art › Fine-Grained Image Classification

Fine-Grained Image Classification

200 papers with code · 36 benchmarks · 41 datasets archive 2025-07-28

Computer Vision

Fine-Grained Image Classification is a task in computer vision where the goal is to classify images into subcategories within a larger category. For example, classifying different species of birds or different types of flowers. This task is considered to be fine-grained because it requires the model to distinguish between subtle differences in visual appearance and patterns, making it more challenging than regular image classification tasks.

( Image credit: Looking for the Devil in the Details )

Description from the archive archive 2025-07-28.

Benchmarks archive 2025-07-28

36 leaderboard tables shown for this task, 36 with rows (a “benchmark” on this site is a table with at least one row, as on /sota), ordered by row count. “Best model” is the first row in the archive's own order at snapshot; nothing is re-ranked here and metric direction is not recorded in the archive. PwC's Trend sparklines are not in the archive, so that column is omitted. 10 shown of 36 until expanded.

Syntology 18 newer papers were judged to report on this task's benchmarks so far (Claude Sonnet 4.5, reading their tables: a model's judgement, not a result; 1 placed): Stanford Cars (1 judged, 0 placed), FGVC Aircraft (2 judged, 0 placed), CUB-200-2011 (2 judged, 1 placed), Oxford 102 Flowers (11 judged, 0 placed), Oxford-IIIT Pets (2 judged, 0 placed), Caltech-101 (1 judged, 0 placed), Food-101 (8 judged, 0 placed), SUN397 (3 judged, 0 placed), MNIST (1 judged, 0 placed). Drawn from 8,886 of 18,637 newer papers checked so far (733 more had nothing to place; 41 could not be read).

DatasetBest model (first row in archive order)PaperCodeSyntologyCompare
Stanford Cars (83 rows) TResnet-L + PMD Progressive Multi-task Anti-Noise Learning and Distilling... code — Compare
FGVC Aircraft (57 rows) I2-HOFI Interweaving Insights: High-Order Feature Interaction for... code — Compare
CUB-200-2011 (30 rows) HERBS Fine-grained Visual Classification with High-temperature... code — Compare
NABirds (30 rows) MetaFormer (MetaFormer-2,384) MetaFormer: A Unified Meta Framework for Fine-Grained Recognition code Syntology official (archive's flag): 9 ran · 9 ran (of which 0 constructed an object rather than computing a result; 8 with no instrument failure: 0 honoured, 1 violated, 7 with no contract checked; 1 where Syntology's instrument failed) · 5 unverified Compare
Oxford 102 Flowers (25 rows) IELT Fine-Grained Visual Classification via Internal Ensemble Learning... code — Compare
Stanford Dogs (24 rows) MP SR-GNN: Spatial Relation-aware Graph Neural Network for... code — Compare
Oxford-IIIT Pets (19 rows) EffNet-L2 (SAM) Sharpness-Aware Minimization for Efficiently Improving Generalization code Syntology official (archive's flag): 7 ran · 15 ran (of which 6 constructed an object rather than computing a result; 13 with no instrument failure: 0 honoured, 0 violated, 13 with no contract checked; 2 where Syntology's instrument failed) · 5 unverified Compare
Caltech-101 (18 rows) VIT-L/16 Reduction of Class Activation Uncertainty with Background Information code — Compare
Food-101 (15 rows) CAP Context-aware Attentional Pooling (CAP) for Fine-grained Visual... code Syntology 0 ran · 2 unverified Compare
Oxford-IIIT Pet Dataset (15 rows) OmniVec2 OmniVec2 - A Novel Transformer based Network for Large Scale... — — Compare
CUB-200-2011 (12 rows) TransFG TransFG: A Transformer Architecture for Fine-grained Recognition code Syntology official (archive's flag): 1 ran · 3 ran (of which 0 constructed an object rather than computing a result; 1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 2 where Syntology's instrument failed) · 0 unverified Compare
CompCars (7 rows) Resnet50 + PMAL Progressive Multi-task Anti-Noise Learning and Distilling... code — Compare
Bird-225 (5 rows) WideResNet-101 (Spinal FC) A Comprehensive Study on Torchvision Pre-trained Models for... code — Compare
Birdsnap (5 rows) EffNet-L2 (SAM) Sharpness-Aware Minimization for Efficiently Improving Generalization code Syntology official (archive's flag): 7 ran · 15 ran (of which 6 constructed an object rather than computing a result; 13 with no instrument failure: 0 honoured, 0 violated, 13 with no contract checked; 2 where Syntology's instrument failed) · 5 unverified Compare
SUN397 (5 rows) µ2Net (ViT-L/16) An Evolutionary Approach to Dynamic Introduction of Tasks in... code — Compare
10 Monkey Species (4 rows) Inception-v3 (Spinal FC) A Comprehensive Study on Torchvision Pre-trained Models for... code — Compare
Fruits-360 (3 rows) ResNeXt-101 A Comprehensive Study on Torchvision Pre-trained Models for... code — Compare
FoodX-251 (2 rows) CSWin-L Learning Multi-Subset of Classes for Fine-Grained Food Recognition code — Compare
Imbalanced CUB-200-2011 (1 row) PC-Softmax Rethinking Softmax with Cross-Entropy: Neural Network Classifier... code — Compare
Kuzushiji-MNIST (1 row) VGG-5 ProgressiveSpinalNet architecture for FC layers code — Compare
BoxCars116K (1 row) ResNet152 + COOC Fine-Grained Vehicle Classification with Unsupervised Parts... — — Compare
iNaturalist (1 row) TASN Looking for the Devil in the Details: Learning Trilinear Attention... code — Compare
Herbarium 2021 Half–Earth (1 row) Conviformer-B Conviformers: Convolutionally guided Vision Transformer code — Compare
Herbarium 2022 (1 row) Conviformer-B Conviformers: Convolutionally guided Vision Transformer code — Compare
Bottles (1 row) PHOC descriptor + Fisher Vector Encoding Fine-grained Image Classification and Retrieval by Combining... code — Compare
CarFlag-1532 (1 row) ResNet101-swp Deep CNNs With Spatially Weighted Pooling for Fine-Grained Car Recognition code — Compare
CarFlag-563 (1 row) ResNet101-swp Deep CNNs With Spatially Weighted Pooling for Fine-Grained Car Recognition code — Compare
Con-Text (1 row) PHOC descriptor + Fisher Vector Encoding Fine-grained Image Classification and Retrieval by Combining... code — Compare
DIB-10K (1 row) MetaFGNet Fine-Grained Visual Categorization using Meta-Learning... code Syntology 7 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 0 honoured, 0 violated, 5 with no contract checked; 2 where Syntology's instrument failed) · 3 unverified Compare
EMNIST-Digits (1 row) VGG-5 ProgressiveSpinalNet architecture for FC layers code — Compare
EMNIST-Letters (1 row) VGG-5 ProgressiveSpinalNet architecture for FC layers code — Compare
FGVC-Aircraft (1 row) EnGraf-Net101 (G=4, H=1) EnGraf-Net: Multiple Granularity Branch Network with Fine-Coarse... code — Compare
MNIST (1 row) Vanilla FC layer only ProgressiveSpinalNet architecture for FC layers code — Compare
QMNIST (1 row) VGG-5 ProgressiveSpinalNet architecture for FC layers code — Compare
SOP (1 row) Assemble-ResNet-FGVC-50 Compounding the Performance Improvements of Assembled Techniques... code Syntology official (archive's flag): 3 ran · 3 ran (of which 0 constructed an object rather than computing a result; 2 with no instrument failure: 0 honoured, 0 violated, 2 with no contract checked; 1 where Syntology's instrument failed) · 8 unverified Compare
STL-10 (1 row) Pre trained wide-resnet-101 ProgressiveSpinalNet architecture for FC layers code — Compare

Syntology column: samples harvested from the paper's repositories and executed on synthesized fixtures; “ran” is not a correctness claim and does not order the table. Each line reads “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the instrument figure counts failures of Syntology's instrument, not of the code. When the archive marks a repository official for the row's paper, the line starts with that repository's state (“official” is the archive's flag, not a verdict on who wrote the code; “community repositories only” when every sample that ran came from a community repository, “official: no sample here; runs from other or unrecorded repositories” when some came from a repository the paper names or has in its text, or from none recorded); hover it for the repositories the samples that ran came from. A dash means no Syntology record for that paper, not a recorded non-run. Each “ran” line links to that paper's sample list. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. For agents: get_harvested_code_for_paper(arxiv_id) lists each paper's samples; how to connect.

Libraries

Not in the archive: the export carries no per-task library table, so there is nothing to show at snapshot 2025-07-28.

Datasets archive 2025-07-28

41 datasets whose archive record lists this task, ordered by the archive's paper count. 30 shown of 41 until expanded.

Subtasks archive 2025-07-28

1 subtask in the archive's task tree.

Parent tasks archive 2025-07-28

Papers archive 2025-07-28

archive papers tagged: 353 · with a code link: 200 · where Syntology ran a sample: 54 (48 with a run with no instrument failure, 6 where every run was a failure of Syntology's instrument) Syntology

Show: all tagged papersonly where code ran (54 of 353 tagged: 48 with a run with no instrument failure, 6 where every run was a failure of Syntology's instrument)

Page 1 of 4: papers 1 to 100 of 353, in archive order: by repositories listed in the archive (most first), then newest first, not by stars (the archive holds no stars, so PwC's “Social” and “Latest” sorts cannot be reproduced). Papers that list no repository come after every paper that lists one.

Papers without a page here are shown as plain text. A Syntology line reads “N ran (of which C constructed an object rather than computing a result; K with no instrument failure: H honoured, V violated, P with no contract checked; I where Syntology's instrument failed) · U unverified”; the figure “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. When the archive marks a repository official for the paper, the line starts with that repository's state (the archive's flag, not a verdict on who wrote the code); hover it for the repositories the samples that ran came from. The first sentence of the abstract is shown for the first 30 papers; every abstract is on its paper's page.

Syntology lines on 46 of the papers shown; no Syntology record for the others (a paper without an arXiv id cannot be joined to the graph, and absence from the graph layer is not a recorded non-run). “Ran” means the sample executed on a synthesized fixture, not that the paper's result was reproduced; each line links to that paper's sample list. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. For agents: get_harvested_code_for_paper(arxiv_id) lists each paper's samples; how to connect.

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