AI Paper Atlas

Research map / Benchmarks / ImageNet val (ResNet-50)

ImageNet val (ResNet-50) leaderboard

12 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on ImageNet val (ResNet-50). Best accuracy: 78.52 (ODConv (4×), ICLR 2022).

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.

ImageNet val (ResNet-50) · accuracy

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

#accuracy ↑MethodPaperTopic
178.52ODConv (4×)
ResNet50
Omni-Dimensional Dynamic Convolution
ICLR 2022 · Table 2
CNN design
278.1EN
ResNet50
Exemplar Normalization for Learning Deep Representation
CVPR 2020 · ImageNet classification result
CNN design
377.5PPBA
ResNet-50
Improving 3D Object Detection through Progressive Population Based Augmentation
ECCV 2020 · Table 6
3D detection and driving
477.1SMOL* 4.1/4
ResNet-50
Not All Bits have Equal Value: Heterogeneous Precisions via Trainable Noise
NeurIPS 2022 · Table 3
Quantization
576.98CE + OPL
ResNet-50
Orthogonal Projection Loss
ICCV 2021 · Table 2
Recognition and long-tail
676.71NIO
ResNet-50
Towards Theoretically Inspired Neural Initialization Optimization
NeurIPS 2022 · Table 4
RNNs and normalization
776Ours + retrain, 3.1 bits
ResNet-50
Transform Quantization for CNN Compression
TPAMI 2021 · Table 2
Quantization
875.79MPN-COV + SEB
ResNet-50
On the Eigenvalues of Global Covariance Pooling for Fine-grained Visual Recognition
TPAMI 2022 · Table 11
Recognition and long-tail
975.75S3 shift 3-bit, random init
ResNet-50
S$^3$: Sign-Sparse-Shift Reparametrization for Effective Training of Low-bit Shift Networks
NeurIPS 2021 · Table 2
Quantization
1074.86RankMixup (M-NDCG)
ResNet-50
RankMixup: Ranking-Based Mixup Training for Network Calibration
ICCV 2023 · Table 2
Recognition and long-tail
2 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

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