代码链接:
https://github.com/Illusionna/ComputerVision/tree/main/EfficientTransformerA repository for ViT. Contribute to Illusionna/Transformer development by creating an account on GitHub.
https://github.com/Illusionna/ComputerVision/tree/main/EfficientTransformer
图片数据集链接:
https://pan.baidu.com/s/1xyMbXGbNYu4nIZRyP50PUg?pwd=emx4
提取码: emx4
执行步骤(先查看 main.py 文档):
可以下载我的权重直接进行分类任务.
先运行 process.py 处理文件.
再运行 train.py 训练文件.
结果:
迭代 700 次准确率接近 99%,测试集基本全部分类正确.