README.md 1.29 KB
Newer Older
sunxx1's avatar
sunxx1 committed
1
# VGG16
sunxx1's avatar
sunxx1 committed
2
3
4

## 模型介绍

sunxx1's avatar
sunxx1 committed
5
VGG模型是2014年ILSVRC竞赛的第二名,第一名是GoogLeNet。但是VGG模型在多个迁移学习任务中的表现要优于GoogLeNet。而且,从图像中提取CNN特征,VGG模型是首选算法。
sunxx1's avatar
sunxx1 committed
6
7
8

## 模型结构

sunxx1's avatar
sunxx1 committed
9
VGG16共有16个层,是一个相当深的卷积神经网络。VGG各种级别的结构都采用了5段卷积,每一段有一个或多个卷积层。
sunxx1's avatar
sunxx1 committed
10
11
12
13
14

## 数据集

在本测试中可以使用ImageNet数据集。

sunxx1's avatar
sunxx1 committed
15
## VGG16训练
sunxx1's avatar
sunxx1 committed
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35

### 环境配置

下载工程代码:

http://developer.hpccube.com/codes/aicomponent/mmclassification-mmcv.git

提供[光源](https://www.sourcefind.cn/#/service-details)拉取的训练的docker镜像:

* 训练镜像:docker pull image.sourcefind.cn:5000/dcu/admin/base/pytorch:1.10.0-centos7.6-dtk-22.10.1-py37-latest
* pip install -r requirements.txt
* cd mmclassification    
* pip install -e .

### 训练

将训练数据解压到data目录下。

训练命令:

sunxx1's avatar
sunxx1 committed
36
    ./vgg16.sh
sunxx1's avatar
sunxx1 committed
37

sunxx1's avatar
sunxx1 committed
38
## 准确率数据
sunxx1's avatar
sunxx1 committed
39
40
41

测试数据使用的是ImageNet数据集,使用的加速卡是DCU Z100L。

sunxx1's avatar
sunxx1 committed
42
43
44
45
46
47
48
| 卡数 |          精度           |
| :--: | :---------------------: |
|  8   | top1:0.7162;top5:0.9049 |

### 历史版本

https://developer.hpccube.com/codes/modelzoo/vgg16-mmcv
sunxx1's avatar
sunxx1 committed
49
50
51
52

### 参考

https://github.com/open-mmlab/mmpretrain