AI Paper Atlas

Research map / Benchmarks / AffNIST

AffNIST leaderboard

10 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on AffNIST. Best accuracy: 98.3 (ET-LP (translation, x-shear), ICML 2019).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

AffNIST · accuracy

10 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
198.3ET-LP (translation, x-shear)
CNN
Equivariant Transformer Networks
ICML 2019 · Additional Experiments table
CNN design
297.2LieLAC [CNN]
CNN
Lie Algebra Canonicalization: Equivariant Neural Operators under arbitrary Lie Groups
ICLR 2025 · MNIST classification table
CNN design
397DTI K-meansDeep Transformation-Invariant Clustering
NeurIPS 2020 · Table 2
Recognition and long-tail
496.7Receptor skeleton (ours)
simple backbone
A Receptor Skeleton for Capsule Neural Networks
ICML 2021 · Table 1
Recognition and long-tail
594.11ConvNet-AvgCapsule Network Is Not More Robust Than Convolutional Network
CVPR 2021 · Table 1
CNN design
693.21Aff-CapsNet + No-R
CapsNet
Improving the Robustness of Capsule Networks to Image Affine Transformations
CVPR 2020 · Table 3
CNN design
792.2SCAEStacked Capsule Autoencoders
NeurIPS 2019 · Table: Ablation study
Open-set and few-shot detection
889.1WholeGroup Equivariant Capsule Networks
NeurIPS 2018 · Table 1
CNN design
984.32ParseCapsParseCaps: An Interpretable Parsing Capsule Network for Medical Image Diagnosis
AAAI 2025 · Table: Robustness analysis on
CNN design
1079CapsNetDynamic Routing Between Capsules
NeurIPS 2017 · text
RNNs and normalization

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.