当前位置：首页 > news >正文

Python实现猫狗分类

news 2025/7/2 3:36:10

不废话了，直接上代码：

def load_imagepath_from_csv(csv_name):image_path = []with open(csv_name,'r') as file:csv_reader = csv.reader(file)next(csv_reader)for row in csv_reader:image_path.append(row[0])return image_pathimport csv
csv_name = "submission_demo.csv"     #文件名仅供参考
image_path_list = load_imagepath_from_csv(csv_name)image_path_listfrom MMEdu import MMClassification as cls
import os
import csv
import numpy as np
rootpath = "test_image"     #文件名仅供参考
csv_name = "submission_demo.csv"     #文件名仅供参考
image_path_list = load_imagepath_from_csv(csv_name)
model = cls(backbone='MobileNet')     #MobileNet也可以换成LeNet,ResNet18,ResNet50,RandForest等
checkpoint = 'checkpoints/cls_model/catsdogs_mobilenet_continue/best_accuracy_top-1_epoch_2.pth'     #文件名仅供参考
predictions = []
for image_name in image_path_list:image_path = rootpath+'/'+image_namey_test_pred = model.inference(image=image_path, show=False, checkpoint=checkpoint, device='cpu')y_test_pred = model.print_result(y_test_pred)predictions.append(y_test_pred)predictionsimport csv
result_csv_path = 'inference_result/results1.csv'
with open(result_csv_path,"w",newline='') as csvfile:csv_writer = csv.writer(csvfile)csv_writer.writerow([f'filename','prediction','pre_class'])for index.image_name in enumerate(image_path_list):csv_writer.writerow([image_name,predictions[index][0]['标签'],predictions[index][0]['预测结果']])csvfile.close()import pandas as pd
df = pd.read_csv(result_csv_path, header=None)
df

下面是数据集（你也可以自己去网上搜图片）：