记录mmaction2模型转onnx

// 转onnx参数
            "args": [
                "configs/skeleton/posec3d/slowonly_r50_u48_240e_ntu120_xsub_keypoint.py",
                "checkpoints/slowonly_r50_u48_240e_ntu120_xsub_keypoint-6736b03f.pth",
                "--verify",
            ]

运行

报错tool/deployment/pytorch2onnx.py
```bash
RuntimeError: Given groups=1, weight of size [32, 17, 1, 7, 7], expected input[1, 1, 3, 224, 224] to have 17 channels, but got 1 channels instead

parser.add_argument(
‘–shape’,
type=int,
nargs=‘+’,
default=[64, 32,17,1,7, 7],

```bash
报错TypeError: forward_dummy() got multiple values for argument 'softmax'

159行

if hasattr(model, 'forward_dummy'):#true #hasattr() 函数用于判断对象是否包含对应的属性
        from functools import partial
        # model.forward = partial(model.forward_dummy, softmax=args.softmax)
        model.forward = model.forward_dummy

修改代码
模型结果tmp.onnx