Skip to content

Latest commit

 

History

History
68 lines (56 loc) · 2.25 KB

model.md

File metadata and controls

68 lines (56 loc) · 2.25 KB

使用模型

mindface人脸套件中包含两类网络结构组件

  • 骨干网络(backbone)：目前支持的有iresnet50, iresnet100, mobilefacenet, vit-tiny, vit-small, vit-base, vit-large等
  • 分类层：PartialFC，用于支持大规模的分类训练

引入现有模型

在人脸识别任务中，常用的需要配置的参数是backbone输出的向量维度，下边的例子的输入图像shape均为(3, 112, 112)，输出维度均为512

import mindspore as ms
import numpy as np
from mindface.recognition.models import *
from mindspore import context

context.set_context(mode=context.PYNATIVE_MODE,
                        device_target="GPU", save_graphs=False)

imgs = np.random.randn(4,3,112,112)
imgs = ms.Tensor(imgs, dtype=ms.float32)

model_name = "iresnet50"

if model_name == "iresnet50":
    model = iresnet50(num_features=512)
elif model_name == "iresnet100":
    model = iresnet100(num_features=512)
elif model_name == "mobilefacenet":
    model = get_mbf(num_features=512)
elif train_info['backbone'] == 'vit_t':
    net = vit_t(num_features=train_info['num_features'])
elif train_info['backbone'] == 'vit_s':
    net = vit_s(num_features=train_info['num_features'])
elif train_info['backbone'] == 'vit_b':
    net = vit_b(num_features=train_info['num_features'])
elif train_info['backbone'] == 'vit_l':
    net = vit_l(num_features=train_info['num_features'])
else:
    raise NotImplementedError

output = model(imgs)
print(output.shape)

fc = PartialFC(num_classes=100, world_size=1)
out = fc(output)
print(out.shape)

添加新模型

  1. 编写新模型的结构 模型的基本结构如下所示
import mindspore.nn as nn

class NewModel(nn.Cell):
    def __init__(self, arg1, arg2):
        super(NewModel, self).__init__()
        pass
    
    def construct(self, data):
        pass

其中__init__函数用初始化模型，包括模型结构搭建和权重初始化，construct执行前向计算。模型的输入为$B,C,H,W$，当前所支持的模型，输入均为三通道($C=3$)，模型的输出为特征向量，作为各类任务head的输入执行下游的计算。

  1. 添加到mindface/recognition/models/init.py中
# 以模型文件名为NewModel为例，在__init__.py中添加引用
from .NewModel import *