MearcAll papers

arXiv 2201.03545 · cs.CV

A ConvNet for the 2020s

Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie

Claims were extracted automatically by reducer-1 on 2026-11-12. Claims are quoted in full under the paper's license.

No reproduction recordedres_6b59515d
Constructed entirely from standard ConvNet modules, ConvNeXts compete favorably with Transformers in terms of accuracy and scalability, achieving 87.8% ImageNet top-1 accuracy and outperforming Swin Transformers on COCO detection and ADE20K segmentation, while maintaining the simplicity and efficiency of standard ConvNets.

Abstract · page 1 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_68d66c32
We demonstrate that a standard ConvNet model can achieve the same level of scalability as hierarchical vision Transformers while being much simpler in design. visual feature learning.

Introduction · page 1 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_1b7ae779
One primary focus of ViT is on the scaling behavior: with the help of larger model and dataset sizes, Transformers can outperform standard ResNets by a significant margin.

Introduction · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_7251d0d5
To do this, we start with a standard ResNet (e.g. ResNet- 50) trained with an improved procedure.

Introduction · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_d7aa1d89
As a result, we propose a family of pure ConvNets dubbed ConvNeXt.

Introduction · page 2 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_88ae884e
In the end, our pure ConvNet model, named ConvNeXt, can outperform the Swin Transformer. follows.

Modernizing a ConvNet: a Roadmap · page 3 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_73814fee
We first train it with similar training techniques used to train vision Transformers and obtain much improved results compared to the original ResNet-50.

Modernizing a ConvNet: a Roadmap · page 3 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_9a60aaec
In Figure 2, we show the procedure and the results we are able to achieve with each step of the “network modernization”.

Modernizing a ConvNet: a Roadmap · page 3 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_55c28489
This improves the model accuracy from 78.8% to 79.4%.

Macro Design · page 3 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_bf5009a4
In this part, we attempt to adopt the idea of ResNeXt [87], which has a better FLOPs/accuracy trade-off than a vanilla ResNet.

ResNeXt-ify · page 4 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_63b8b4d2
Despite the increased FLOPs for the depthwise convolution layer, this change reduces the whole network FLOPs to 4.6G, due to the significant FLOPs reduction in the downsampling residual blocks’ shortcut 1×1 conv layer.

Inverted Bottleneck · page 4 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_874c8a72
Interestingly, this results in slightly improved performance (80.5% to 80.6%).

Inverted Bottleneck · page 4 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_d5abc08d
This intermediate step reduces the FLOPs to 4.1G, resulting in a temporary performance degradation to 79.9%.

Large Kernel Sizes · page 4 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_5b3e0dd1
Additionally, we observe that the benefit of larger kernel sizes reaches a saturation point at 7×7.

Large Kernel Sizes · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_d7e0c920
This process improves the result by 0.7% to 81.3%, practically matching the performance of Swin-T.

Micro Design · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_bf10dccb
This further boosts the performance to 81.4%, already surpassing Swin-T’s result.

Micro Design · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_dd583349
For simplicity, we note the linear layers in Transformer MLP blocks also as “1×1 convs” since they are equivalent. that we have even fewer normalization layers per block than Transformers, as empirically we find that adding one additional BN layer at the beginning of the block does not improve the performance.

Micro Design · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_12807b5c
We observe that our ConvNet model does not have any difficulties training with LN; in fact, the performance is slightly better, obtaining an accuracy of 81.5%.

Micro Design · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_aed3b8a3
In ResNet, the spatial downsampling is achieved by the residual block at the start of each stage, using 3×3 conv with stride 2 (and 1×1 conv with stride 2 at the shortcut connection).

Micro Design · page 5 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_51e6db2b
We can improve the accuracy to 82.0%, significantly exceeding Swin-T’s 81.3%.

Micro Design · page 6 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_0f62d6e2
We have finished our first “playthrough” and discovered ConvNeXt, a pure ConvNet, that can outperform the Swin Transformer for ImageNet-1K classification in this compute regime.

Micro Design · page 6 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_42838c93
ConvNeXt also outperforms Swin Transformer of similar complexities across the model imagesize #param.

Results · page 6 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_ae3fbebf
A highlight from the results is ConvNeXt-B at 384 2: it outperforms Swin-B by 0.6% (85.1% vs. 84.5%), but with 12.5% higher inference throughput (95.7 vs. 85.1 image/s).

Results · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_add1c4c0
Additionally, we observe an improved result of 85.5% when further scaling to ConvNeXt-L.

Results · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_fee08983
We present results with models fine-tuned from ImageNet-22K pre-training at Table 1 (lower).

Results · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_4f3ebb98
Our results demonstrate that properly designed ConvNets are not inferior to vision Transformers when pre-trained with large dataset — ConvNeXts still perform on par or better than similarly-sized Swin Transformers, with slightly higher throughput.

Results · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_08ab5cfb
Additionally, our ConvNeXt-XL model achieves an accuracy of 87.8% — a decent improvement over ConvNeXt-L at 3842, demonstrating that ConvNeXts are scalable architectures.

Results · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_4bc4c0eb
However, with ImageNet-22K pre-training, ConvNeXt is able to outperform EfficientNetV2, further demonstrating the importance of large-scale training.

Results · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_266997e6
We use the supervised training results from DeiT [73] for ViT-S/B and MAE [26] for ViT-L, as they employ improved training procedures over the original ViTs [20].

Results · page 7 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_3aa05c16
When scaled up to bigger models (ConvNeXt-B/L/XL) pre-trained on ImageNet-22K, in many cases ConvNeXt is significantly better(e.g. +1.0 AP) than Swin Transformers in terms of box and mask AP.

Empirical Evaluation on Downstream Tasks · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_559c2424
This is true for both classification and other tasks requiring higher-resolution inputs (see Table 1,3 for comparisons of throughput/FPS).

Empirical Evaluation on Downstream Tasks · page 8 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedclm_986b25ba
We propose ConvNeXts, a pure ConvNet model that can compete favorably with state-of-the-art hierarchical vision Transformers across multiple computer vision benchmarks, while retaining the simplicity and efficiency of standard ConvNets.

Conclusions · page 9 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_51d2595a
The only exception is ConvNeXt-L pre-trained on ImageNet-1K, where the model accuracy is significantly lower than the EMA accuracy due to overfitting, and we select its best EMA model during pre-training as the starting point for fine-tuning.

ImageNet Fine-tuning · page 9 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_111895b9
Additional robustness evaluation results for ConvNeXt models are presented in Table 8.

Robustness Evaluation · page 10 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_0562a396
We directly test our ImageNet-1K trained/fine-tuned classification models on several robustness benchmark datasets such as ImageNet-A [33], ImageNet-R [30], ImageNet-Sketch [78] and ImageNet- C/¯C [31, 48] datasets.

Robustness Evaluation · page 10 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_9e304d26
With extra ImageNet-22K data, ConvNeXt- XL demonstrates strong domain generalization capabilities (e.g. achieving 69.3%/68.2%/55.0% accuracy on ImageNet- A/R/Sketch benchmarks, respectively).

Robustness Evaluation · page 10 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_0a21fe6c
One interesting difference is that inverting dimensions brings a larger improvement at ResNet-200 regime than at ResNet-50 regime (+0.79% vs. +0.14%).

Modernizing ResNets: detailed results · page 10 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_593a6295
Using fewer normalization layers also has a bigger gain compared with the ResNet-50 regime (+0.46% vs. +0.14%). model IN-1K acc.

Modernizing ResNets: detailed results · page 11 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_ef3bcb52
Swin Transformers and ConvNeXts both achieve faster inference throughput than V100 GPUs, but ConvNeXts’ advantage is now significantly greater, sometimes up to 49% faster.

Benchmarking on A100 GPUs · page 12 · Automated extraction, not reviewed · 0 reviewed reproductions

No reproduction recordedres_7dc3c277
Using TF32 data format and “channel last” memory layout, ConvNeXt enjoys up to ∼49% higher throughput compared with a Swin Transformer with similar FLOPs.

Benchmarking on A100 GPUs · page 12 · Automated extraction, not reviewed · 0 reviewed reproductions

Reproduced one of these claims?

Record your result in the Mearc app on your own machine, then use Publish this result with this paper's id (2201.03545) and the claim id shown next to the claim. Mearc creates a file that holds only what you choose to publish. Add it to the public claim data by pull request; a maintainer other than you reviews it before it counts.