AI Paper Atlas

Research map / Benchmarks / ImageNet (ILSVRC2012)

ImageNet (ILSVRC2012) leaderboard

20 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on ImageNet (ILSVRC2012). Best accuracy: 77.2 (FBNetV2-L1, CVPR 2020).

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 (ILSVRC2012) · accuracy

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

#accuracy ↑MethodPaperTopic
177.2FBNetV2-L1FBNetV2: Differentiable Neural Architecture Search for Spatial and Channel Dimensions
CVPR 2020 · ImageNet classification perfor
Neural architecture search
277HourNAS-EfficientNetSS-FHourNAS: Extremely Fast Neural Architecture Search Through an Hourglass Lens
CVPR 2021 · Overall comparison table
Neural architecture search
376.8FGD
ResNet50
Feedback Gradient Descent: Efficient and Stable Optimization with Orthogonality for DNNs
AAAI 2022 · Table: Test accuracy rates on
Optimizers
476.6One-Shot-NAS-GCN†Fitting the Search Space of Weight-sharing NAS with Graph Convolutional Networks
AAAI 2021 · Table: Comparison with state-o
Neural architecture search
576.11MetaDistiller
ResNet50
MetaDistiller: Network Self-Boosting via Meta-Learned Top-Down Distillation
ECCV 2020 · Table 2
Knowledge distillation
675.7GhostNet 1.3×GhostNet: More Features From Cheap Operations
CVPR 2020 · Table: Comparison of state-of-
CNN design
774.02DST
ResNet-50
Dynamic Sparse Training: Find Efficient Sparse Network From Scratch With Trainable Masked Layers
ICLR 2020 · Table 3
Pruning and efficient tuning
872.6PeleeNetPelee: A Real-Time Object Detection System on Mobile Devices
NeurIPS 2018 · ImageNet results table
CNN features
964.6HWGQ-Net
ResNet-50
Deep Learning With Low Precision by Half-Wave Gaussian Quantization
CVPR 2017 · Table: The results of various
Quantization
1057.9CLIP-Q
AlexNet
CLIP-Q: Deep Network Compression Learning by In-Parallel Pruning-Quantization
CVPR 2018 · Table 1
Pruning and efficient tuning

ImageNet (ILSVRC2012) · accuracy (val)

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

#accuracy ↑MethodPaperTopic
190.2Meta Pseudo Labels (EfficientNet-L2)
EfficientNet-L2
Meta Pseudo Labels
CVPR 2021 · Table: Large Scale Experiment
Semi-supervised learning
287.3EfficientNetV2-XL (21k)EfficientNetV2: Smaller Models and Faster Training
ICML 2021 · Table: EfficientNetV2 Performa
Neural architecture search
379SpineNet-143SpineNet: Learning Scale-Permuted Backbone for Recognition and Localization
CVPR 2020 · Image classification results t
CNN features
476.32SCAN Classifier 4/4
ResNet101
SCAN: A Scalable Neural Networks Framework Towards Compact and Efficient Models
NeurIPS 2019 · Table: Experiments results of
RNNs and normalization
574.62TNet, 5 locations
BagNet-77
Hard-Attention for Scalable Image Classification
NeurIPS 2021 · Effectiveness table
Attention and transformers
670.5ChannelNet-v1ChannelNets: Compact and Efficient Convolutional Neural Networks via Channel-Wise Convolutions
NeurIPS 2018 · Table: Comparison of ChannelNe
CNN design
769.6ReActNet-A (Diode)
ReActNet-A
Enhancing Optimization Robustness in 1-bit Neural Networks through Stochastic Sign Descent
ECCV 2024 · Table 1
Optimizers
867.8Sketch (ref.)
ResNet-18 type B
Network Sketching: Exploiting Binary Structure in Deep CNNs
CVPR 2017 · Table 5
CNN design
956Newton type preconditioned SGD
AlexNet
Preconditioner on Matrix Lie Group for SGD
ICLR 2019 · text
Optimizers
1055.3DA-CNN
AlexNet
Growing a Brain: Fine-Tuning by Increasing Model Capacity
CVPR 2017 · Table 4
Pruning and efficient tuning

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