不用C++编译、CUDA配置什么的，我改了一下推理的代码，可以直接跑，但

[zhaoyun0071]

将"preprocess\humanparsing\modules\functions.py"中的内容替换如下，可以跑起来。

from os import path

import torch.distributed as dist
import torch.autograd as autograd

_src_path = path.join(path.dirname(path.abspath(__file__)), "src")


# Activation names
ACT_RELU = "relu"
ACT_LEAKY_RELU = "leaky_relu"
ACT_ELU = "elu"
ACT_NONE = "none"
def normalize_shape(x):
    if x.ndimension() == 1:
        return x.view(1, -1, 1)
    else:
        return x.view(x.size(0), x.size(1), -1)

import torch


def broadcast_to(v, x):
    if x.ndimension() == 2:
        return v
    else:
        broadcast_size = [1, -1]
        for i in range(2, x.ndimension()):
            broadcast_size.append(1)

        return v.view(broadcast_size)


def forward_cpu(x, mean, var, weight, bias, affine, eps):
    gamma = torch.abs(weight) + eps if affine else torch.ones_like(var)
    mul = torch.rsqrt(var + eps) * gamma

    x.sub_(broadcast_to(mean, x))
    x.mul_(broadcast_to(mul, x))
    if affine:
        x.add_(broadcast_to(bias, x))

    return x

def _check(fn, *args, **kwargs):
    success = fn(*args, **kwargs)
    if not success:
        raise RuntimeError("CUDA Error encountered in {}".format(fn))




def _reduce(x):
    if len(x.size()) == 2:
        return x.sum(dim=0)
    else:
        n, c = x.size()[0:2]
        return x.contiguous().view((n, c, -1)).sum(2).sum(0)


def _count_samples(x):
    count = 1
    for i, s in enumerate(x.size()):
        if i != 1:
            count *= s
    return count
def leaky_relu_forward(z, slope):

    return torch.nn.functional.leaky_relu(z, negative_slope=slope)

def _act_forward(ctx, x):
    if ctx.activation == ACT_LEAKY_RELU:
        leaky_relu_forward(x, ctx.slope)



class InPlaceABN(autograd.Function):
    @staticmethod
    def forward(ctx, x, weight, bias, running_mean, running_var,
                training=False, momentum=0.1, eps=1e-05, activation=ACT_LEAKY_RELU, slope=0.01):
        # Save context
        ctx.training = training
        ctx.momentum = momentum
        ctx.eps = eps
        ctx.activation = activation
        ctx.slope = slope
        ctx.affine = weight is not None and bias is not None

        # Prepare inputs
        count = _count_samples(x)
        x = x.contiguous()
        weight = weight.contiguous() if ctx.affine else x.new_empty(0)
        bias = bias.contiguous() if ctx.affine else x.new_empty(0)


        mean, var = running_mean.contiguous(), running_var.contiguous()
        ctx.mark_dirty(x)

        # BN forward + activation
        forward_cpu(x, mean, var, weight, bias, ctx.affine,ctx.eps)
        _act_forward(ctx, x)

        # Output
        ctx.var = var
        ctx.save_for_backward(x, var, weight, bias)
        ctx.mark_non_differentiable(running_mean, running_var)
        return x, running_mean, running_var


class InPlaceABNSync(autograd.Function):
    @classmethod
    def forward(cls, ctx, x, weight, bias, running_mean, running_var,
                training=False, momentum=0.1, eps=1e-05, activation=ACT_LEAKY_RELU, slope=0.01, equal_batches=True):
        # Save context
        ctx.training = training
        ctx.momentum = momentum
        ctx.eps = eps
        ctx.activation = activation
        ctx.slope = slope
        ctx.affine = weight is not None and bias is not None

        # Prepare inputs
        ctx.world_size = dist.get_world_size() if dist.is_initialized() else 1

        # count = _count_samples(x)
        batch_size = x.new_tensor([x.shape[0]], dtype=torch.long)

        x = x.contiguous()
        weight = weight.contiguous() if ctx.affine else x.new_empty(0)
        bias = bias.contiguous() if ctx.affine else x.new_empty(0)


        mean, var = running_mean.contiguous(), running_var.contiguous()
        ctx.mark_dirty(x)

        # BN forward + activation
        forward_cpu(x, mean, var, weight, bias, ctx.affine,ctx.eps)
        _act_forward(ctx, x)

        # Output
        ctx.var = var
        ctx.save_for_backward(x, var, weight, bias)
        ctx.mark_non_differentiable(running_mean, running_var)
        return x, running_mean, running_var



inplace_abn = InPlaceABN.apply
inplace_abn_sync = InPlaceABNSync.apply

__all__ = ["inplace_abn", "inplace_abn_sync", "ACT_RELU", "ACT_LEAKY_RELU", "ACT_ELU", "ACT_NONE"]

但，衣服覆盖的不全，这是为什么呢。 forward_cpu代码是"\preprocess\humanparsing\modules\src\inplace_abn_cpu.cpp"中的C++代码转写成python的