ResNet

Category: CNN · Code: timm/models/resnet.py · Architectures: 81 · Pretrained weights: 149 (144 with ImageNet-1k results)

Residual networks (ResNet) and the variants built on the same ResNet class: ResNeXt grouped convolutions, Squeeze-and-Excitation (SE) and ECA channel attention, deep / tiered stems and average pool downsampling (the “c”, “d”, “s” and “t” variants), anti-aliased downsampling, Wide ResNet and ResNet-RS configurations.

Papers

• Deep Residual Learning for Image Recognition
• Bag of Tricks for Image Classification with Convolutional Neural Networks
• Aggregated Residual Transformations for Deep Neural Networks
• Wide Residual Networks
• Squeeze-and-Excitation Networks
• ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks
• Making Convolutional Networks Shift-Invariant Again
• Group Normalization
• Revisiting ResNets: Improved Training and Scaling Strategies
• ResNet strikes back: An improved training procedure in timm
• Billion-scale semi-supervised learning for image classification
• Exploring the Limits of Weakly Supervised Pretraining
• Knapsack Pruning with Inner Distillation

Usage

See the Quickstart for loading pretrained weights, image preprocessing and fine-tuning, and Feature Extraction for using the models as backbones.

>>> import timm
>>> model = timm.create_model('resnet50.a1_in1k', pretrained=True)
>>> data_config = timm.data.resolve_model_data_config(model)

Model API support

✓ supported, ✓ (square) works for square sizes only, ✗ not supported, – not needed (the model already accepts any input size). Checked by running each feature on a randomly initialized model (ResNet with resnet10t at 176), other architectures of the same class are expected to behave the same. See the Model API page for details.

Architectures

GMACs and parameter counts are for the default input size and classifier of each architecture, from the benchmark results.

Pretrained weights

Top-1 is ImageNet-1k validation accuracy from results-imagenet.csv at the listed eval size, - if the weight has no ImageNet-1k result (e.g. pretraining or image-text weights). Classes is the size of the pretrained classifier, none for weights without one.

Weights are grouped by source: the paper or origin given in the pretrained config, else a Hub organization other than timm, else the first part of the pretrained tag (architecture.tag). Sources with a single weight are listed under other.

ResNet strikes back

Pretrained configs reference ResNet strikes back: An improved training procedure in timm.

huggingface/pytorch-image-models

Origin: github.com/huggingface/pytorch-image-models

cv.gluon.ai

Origin: cv.gluon.ai/model_zoo/classification.html

pytorch/vision

Origin: github.com/pytorch/vision

facebookresearch/semi-supervised-ImageNet1K-models

Origin: github.com/facebookresearch/semi-supervised-ImageNet1K-models

tf tags

miil tags

facebookresearch/WSL-Images

Origin: github.com/facebookresearch/WSL-Images

sw tags

Other