AI Tools.

Search

image classification by timm

edgenext_small.usi_in1k

EdgeNeXt-Small is a lightweight CNN-transformer hybrid architecture optimized for mobile and edge inference, pre-trained on ImageNet-1K with Universal Self-Attention Interaction (USI) training. MIT-licensed and available via timm's model registry.

Summary text generated by an automated pipeline from the model card · Not individually reviewed or run by us · How this page is made

From the model card

Fields below are copied from the tags and counters on the HuggingFace repository timm/edgenext_small.usi_in1k at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.

Publisher (HF namespace)
timm
Pipeline tag
image-classification
Library
timm, Transformers
Framework tags
PyTorch
Weight formats
safetensors
License tag
mit — read the license file in the repo before relying on it
Papers cited
arXiv:2206.10589, arXiv:2204.03475
Datasets declared
imagenet-1k
Downloads (HF counter at last fetch)
361,500
Likes (HF counter at last fetch)
6
Model card
https://huggingface.co/timm/edgenext_small.usi_in1k

Use cases

  • Mobile image classification with hybrid CNN-transformer efficiency
  • Edge device deployment requiring fast, low-power inference
  • Backbone for lightweight object detection in embedded systems
  • Ablation study on CNN-transformer architecture tradeoffs

Pros

  • MIT license
  • Hybrid architecture balances accuracy and mobile inference efficiency
  • ImageNet-1K pretrained — broad visual transfer
  • timm registry integration for easy fine-tuning

Cons

  • timm dependency required — not native Transformers pipeline
  • Small scale trades accuracy for speed — not competitive on challenging benchmarks
  • USI training methodology is less documented than standard distillation
  • Limited community fine-tuning examples compared to ViT or ConvNeXt families

Tags

timmpytorchsafetensorsimage-classificationtransformersdataset:imagenet-1karxiv:2206.10589arxiv:2204.03475license:mitregion:us