facebookresearch/ConvNeXt

Health v2 · maintenance only

Flags: no_releases archived

Adoption not part of the score

6416 stars · 745 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Official PyTorch implementation of ConvNeXt, a pure convolutional neural network architecture from the CVPR 2022 paper 'A ConvNet for the 2020s'. It includes training, fine-tuning, and downstream transfer code plus pre-trained ImageNet-1K and ImageNet-22K model checkpoints.

Use cases

• classify images with a modern pure ConvNet
• train ConvNeXt on ImageNet-1K
• fine-tune ConvNeXt on a custom dataset like CIFAR
• use ConvNeXt as a backbone for object detection or segmentation
• download pre-trained ConvNeXt checkpoints for transfer learning
• compare ConvNet vs Vision Transformer performance

When to choose

• you need a strong, efficient convolutional image classification backbone
• you want pre-trained ImageNet models for detection or segmentation transfer
• you prefer ConvNet architectures over transformers for vision tasks

When to avoid

• you need a Vision Transformer instead of a ConvNet
• you want a maintained production inference library rather than research code
• you need non-PyTorch frameworks like TensorFlow or JAX

Facets

library · maturity maintenance

machine-learning deep-learning image-processing computer-vision deep-learning computer-vision image-processing machine-learning python convnext convnet image-classification pytorch pretrained-models research-code facebook-research gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="facebookresearch/ConvNeXt")