ResNeXt is a classification network further developed after ResNet, which combines stacking blocks and split-transform-merge.
| Model | Download | Top-1 accuracy (%) | Top-5 accuracy (%) |
|---|---|---|---|
| ResNeXt | 340M | 79.316 | 94.518 |
- ResNeXt model was described in the paper titled Aggregated Residual Transformations for Deep Neural Networks
- The pre-train model is from the Project: pytorch-image-models.
Apache 2.0