Skip to content

Latest commit

 

History

History
57 lines (49 loc) · 4.57 KB

repvgg.md

File metadata and controls

57 lines (49 loc) · 4.57 KB
Raw

RepVGG

  • Paper：RepVGG: Making VGG-style ConvNets Great Again

  • Origin Repo：DingXiaoH/RepVGG

  • Code：repvgg.py

  • Evaluate Transforms：

    # backend: pil
# input_size: 224x224
transforms = T.Compose([
    T.Resize(256),
    T.CenterCrop(224),
    T.ToTensor(),
    T.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])

  • Model Details：

    Model Model Name Params (M) FLOPs (G) Top-1 (%) Top-5 (%) Pretrained Model
    RepVGG-A0 repvgg_a0 8.3 1.4 72.42 90.49 Download
    RepVGG-A1 repvgg_a1 12.8 2.4 74.46 91.85 Download
    RepVGG-A2 repvgg_a2 25.5 5.1 76.46 93.00 Download
    RepVGG-B0 repvgg_b0 14.3 3.1 75.15 92.42 Download
    RepVGG-B1 repvgg_b1 51.8 11.8 78.37 94.10 Download
    RepVGG-B2 repvgg_b2 80.3 18.4 78.79 94.42 Download
    RepVGG-B3 repvgg_b3 111.0 26.2 80.50 95.26 Download
    RepVGG-B1g2 repvgg_b1g2 41.4 8.8 77.80 93.88 Download
    RepVGG-B1g4 repvgg_b1g4 36.1 7.3 77.58 93.84 Download
    RepVGG-B2g4 repvgg_b2g4 55.8 11.3 79.38 94.68 Download
    RepVGG-B3g4 repvgg_b3g4 75.6 16.1 80.21 95.09 Download

  • Citation：

    @article{ding2021repvgg,
    title={RepVGG: Making VGG-style ConvNets Great Again},
    author={Ding, Xiaohan and Zhang, Xiangyu and Ma, Ningning and Han, Jungong and Ding, Guiguang and Sun, Jian},
    journal={arXiv preprint arXiv:2101.03697},
    year={2021}
}