osmr/imgclsmobGitHub
Stale multi-framework image classification model zoo; choose if you need Keras/TFLite models, not PyTorch-only upkeep.
Sandbox for training deep learning networks
- 3.0k
- Python
- MIT
dormant2.1y ago
huggingface/pytorch-image-models Alternatives
image classification / vision backbones
A PyTorch library providing a large collection of image classification models and vision backbones (ResNet, EfficientNet, ViT, etc.) with pretrained weights and training/eval/inference/export scripts.
Older generation, or narrower
An earlier or more limited way to do the job. Still the right call when you need something small, proven, or CPU-only.
apple/ml-cvnetsGitHub
Apple's config-driven CV training library; choose for its integrated pipelines, but stale and less focused than timm.
CVNets: A library for training computer vision networks
- 2.0k
- Python
- NOASSERTION
active29d ago
rwightman/gen-efficientnet-pytorchGitHub
EfficientNet-family-only predecessor of timm; choose for a minimal dependency on just those architectures.
Pretrained EfficientNet, EfficientNet-Lite, MixNet, MobileNetV3 / V2, MNASNet A1 and B1, FBNet, Single-Path NAS
- 1.6k
- Python
- Apache-2.0
dormant2.3y ago
lucidrains/vit-pytorchGitHub
Standalone ViT variant implementations without pretrained weights; choose for research flexibility but not drop-in.
Implementation of Vision Transformer, a simple way to achieve SOTA in vision classification with only a single transformer encoder, in Pytorch
- 25.5k
- Python
- MIT
active20d ago
Cadene/pretrained-models.pytorchGitHub
Older pretrained ConvNet zoo; choose if you need its original weight APIs, but it is unmaintained.
Pretrained ConvNets for pytorch: NASNet, ResNeXt, ResNet, InceptionV4, InceptionResnetV2, Xception, DPN, etc.
- 9.1k
- Python
- BSD-3-Clause
dormant4.5y ago
NVlabs/FANGitHub
Narrow implementation of Fully Attentional Networks; choose if you need that exact architecture/weights, but stale and not a general zoo.
Official PyTorch implementation of Fully Attentional Networks
- 483
- Python
- NOASSERTION
dormant3.5y ago
Listed as an alternative to
These projects were analysed and named pytorch-image-models among their alternatives. The relationship is not symmetric — how pytorch-image-models rates them is a separate judgement, made when pytorch-image-models is analysed in its own right.
microsoft/Swin-TransformerGitHub
calls pytorch-image-models “Choose it when you want an actively maintained single PyTorch library that includes Swin plus many other pretrained backbones.”
This is an official implementation for "Swin Transformer: Hierarchical Vision Transformer using Shifted Windows".
16.1kPythondormant
rwightman/gen-efficientnet-pytorchGitHub
calls pytorch-image-models “Choose it as the actively maintained superset that the seed's readme recommends replacing it with.”
Pretrained EfficientNet, EfficientNet-Lite, MixNet, MobileNetV3 / V2, MNASNet A1 and B1, FBNet, Single-Path NAS
1.6kPythondormant
WZMIAOMIAO/deep-learning-for-image-processingGitHub
calls pytorch-image-models “Use when you need production-ready pretrained image encoders and training scripts instead of hand-written tutorial code.”
deep learning for image processing including classification and object-detection etc.
26.4kPythonslowing
lucidrains/vit-pytorchGitHub
calls pytorch-image-models “Large model collection; choose it when you want pretrained weights and standardized training/inference across many backbones.”
Implementation of Vision Transformer, a simple way to achieve SOTA in vision classification with only a single transformer encoder, in Pytorch
25.5kPythonactive
NVIDIA/DeepLearningExamplesGitHub
calls pytorch-image-models “choose when you want a maintained PyTorch library with hundreds of pretrained image backbones rather than standalone NVIDIA example scripts.”
State-of-the-Art Deep Learning scripts organized by models - easy to train and deploy with reproducible accuracy and performance on enterprise-grade infrastructure.
14.8kJupyter Notebookdormant
google-research/vision_transformerGitHub
calls pytorch-image-models “choose when you need ViT/MLP-Mixer or many other backbones in PyTorch with pretrained weights instead of JAX code.”
12.7kJupyter Notebookactive