用torchvision的例子进行分类预测试验，所用数据集为flowers102，所用模型为Resnet152

1。下载地址：

https://www.robots.ox.ac.uk/~vgg/data/flowers/102/

image.png

2。训练测试集分类代码处理：

2.1 下载数据如红框所示，解压后的数据在jpg文件夹中，未对训练，验证集，测试集进行划分；

image.png

2.2 用如下代码进行train, valid, test集划分：

import scipy.io
import numpy as np
import os
from PIL import Image
import shutil

labels = scipy.io.loadmat('./flower_data/flowers-102/imagelabels.mat')#该地址为imagelabels.mat的绝对地址
labels = np.array(labels['labels'][0]) - 1
print("labels:", labels)
setid = scipy.io.loadmat('./flower_data/flowers-102/setid.mat')#该地址为setid.mat的绝对地址
validation = np.array(setid['valid'][0]) - 1
np.random.shuffle(validation)
train = np.array(setid['trnid'][0]) - 1
np.random.shuffle(train)
test = np.array(setid['tstid'][0]) - 1
np.random.shuffle(test)
flower_dir = list()

for img in os.listdir("./flower_data/flowers-102/jpg"):#该地址为源数据图片的绝对地址
    flower_dir.append(os.path.join("./flower_data/flowers-102/jpg", img))
flower_dir.sort()
print(flower_dir)

des_folder_train = "./flower_data/train"#该地址为新建的训练数据集文件夹的绝对地址

for tid in train:#打开图片并获取标签
    img = Image.open(flower_dir[tid])
    # print(img)
    # print(flower_dir[tid])
    img = img.resize((256, 256), Image.ANTIALIAS)
    lable = labels[tid]
    # print(lable)

    path = flower_dir[tid]
    # print("path:", path)
    base_path = os.path.basename(path)
    # print("base_path:", base_path)
    classes = "c" + str(lable)
    class_path = os.path.join(des_folder_train, classes)
    # 判断结果
    if not os.path.exists(class_path):
        os.makedirs(class_path)
    # print("class_path:", class_path)
    despath = os.path.join(class_path, base_path)
    # print("despath:", despath)
    img.save(despath)

des_folder_validation = "./flower_data/valid"#该地址为新建的验证数据集文件夹的绝对地址

for tid in validation:
    img = Image.open(flower_dir[tid])
    # print(flower_dir[tid])
    img = img.resize((256, 256), Image.ANTIALIAS)
    lable = labels[tid]
    # print(lable)
    path = flower_dir[tid]
    # print("path:", path)
    base_path = os.path.basename(path)
    # print("base_path:", base_path)
    classes = "c" + str(lable)
    class_path = os.path.join(des_folder_validation, classes)
    # 判断结果
    if not os.path.exists(class_path):
        os.makedirs(class_path)
    # print("class_path:", class_path)
    despath = os.path.join(class_path, base_path)
    # print("despath:", despath)
    img.save(despath)

des_folder_test = "./flower_data/test"#该地址为新建的测试数据集文件夹的绝对地址

for tid in test:
    img = Image.open(flower_dir[tid])
    # print(flower_dir[tid])
    img = img.resize((256, 256), Image.ANTIALIAS)
    lable = labels[tid]
    # print(lable)
    path = flower_dir[tid]
    # print("path:", path)
    base_path = os.path.basename(path)
    # print("base_path:", base_path)
    classes = "c" + str(lable)
    class_path = os.path.join(des_folder_test, classes)
    # 判断结果
    if not os.path.exists(class_path):
        os.makedirs(class_path)
    # print("class_path:", class_path)
    despath = os.path.join(class_path, base_path)
    # print("despath:", despath)
    img.save(despath)

2.3 划分后的形式如下，完成模型所需数据格式。

image.png

image.png