2020-04-15 18:54:00 +08:00
|
|
|
|
# 数据说明
|
|
|
|
|
|
|
|
|
|
|
|
---
|
|
|
|
|
|
|
|
|
|
|
|
## 1.简介
|
|
|
|
|
|
PaddleClas支持ImageNet1000和Flower数据分类任务。
|
|
|
|
|
|
PaddleClas提供了丰富的预训练模型，支持的模型列表请参考[模型库](../models/models_intro.md)
|
|
|
|
|
|
|
|
|
|
|
|
## 2.数据集准备
|
|
|
|
|
|
|
|
|
|
|
|
数据集 | 训练集大小 | 测试集大小 | 类别数 | 备注|
|
|
|
|
|
|
:------:|:---------------:|:---------------------:|:-----------:|:-----------:
|
|
|
|
|
|
Flowers|1k | 6k | 102 |
|
|
|
|
|
|
[ImageNet](http://www.image-net.org/challenges/LSVRC/2012/)|1.2M| 50k | 1000 |
|
|
|
|
|
|
|
|
|
|
|
|
数据格式
|
|
|
|
|
|
|
|
|
|
|
|
PaddleClas加载PaddleClas/dataset/中的数据，请将下载后的数据按下面格式组织放置到PaddleClas/dataset/中。
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
PaddleClas/dataset/imagenet
|
|
|
|
|
|
|_ train
|
|
|
|
|
|
| |_ n01440764
|
|
|
|
|
|
| | |_ n01440764_10026.JPEG
|
|
|
|
|
|
| | |_ ...
|
|
|
|
|
|
| |_ ...
|
|
|
|
|
|
| |
|
|
|
|
|
|
| |_ n15075141
|
|
|
|
|
|
| |_ ...
|
|
|
|
|
|
| |_ n15075141_9993.JPEG
|
|
|
|
|
|
|_ val
|
|
|
|
|
|
| |_ ILSVRC2012_val_00000001.JPEG
|
|
|
|
|
|
| |_ ...
|
|
|
|
|
|
| |_ ILSVRC2012_val_00050000.JPEG
|
|
|
|
|
|
|_ train_list.txt
|
|
|
|
|
|
|_ val_list.txt
|
2020-04-15 18:58:10 +08:00
|
|
|
|
```
|
2020-04-15 18:54:00 +08:00
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
PaddleClas/dataset/flower
|
|
|
|
|
|
|_ train
|
|
|
|
|
|
| |_ image_03601.jpg
|
|
|
|
|
|
| |_ ...
|
|
|
|
|
|
| |_ image_07073.jpg
|
|
|
|
|
|
|_ val
|
|
|
|
|
|
| |_ image_04121.jpg
|
|
|
|
|
|
| |_ ...
|
|
|
|
|
|
| |_ image_02355.jpg
|
|
|
|
|
|
|_ train_list.txt
|
|
|
|
|
|
|_ val_list.txt
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
或是通过软链接将数据从实际地址链接到PaddleClas/dataset/下
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
#imagenet
|
|
|
|
|
|
ln -s actual_path/imagenet path_to_PaddleClas/dataset/imagenet
|
|
|
|
|
|
|
|
|
|
|
|
#flower
|
|
|
|
|
|
ln -s actual_path/flower path_to_PaddleClas/dataset/flower
|
|
|
|
|
|
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
## 3.下载预训练模型
|
|
|
|
|
|
通过tools/download.py下载所需要的预训练模型。
|
|
|
|
|
|
|
|
|
|
|
|
```bash
|
|
|
|
|
|
python tools/download.py -a ResNet50_vd -p ./pretrained -d True
|
|
|
|
|
|
```
|
|
|
|
|
|
|
|
|
|
|
|
参数说明：
|
|
|
|
|
|
+ `architecture`（简写 a）：模型结构
|
|
|
|
|
|
+ `path`（简写 p）：下载路径
|
|
|
|
|
|
+ `decompress` （简写 d）：是否解压
|
