15、深度学习：从肺炎预测到目标检测

深度学习：从肺炎预测到目标检测

1. 肺炎预测

在计算机视觉领域，卷积神经网络（CNN）是一种强大的算法。下面将介绍如何使用预先训练好的CNN模型对新的图像集进行肺炎预测。

1.1 代码实现

以下是使用训练好的CNN模型预测肺炎的代码：

import numpy as np
import pathlib
import cv2
import tensorflow as tf
import matplotlib.pyplot as plt

model_path = "models/pneumiacnn"
val_img_dir = "images/chest_xray/val"
# ImageDataGenerator class provides a mechanism to load both small and large dataset.
# Instruct ImageDataGenerator to scale to normalize pixel values to range (0, 1)
datagen = tf.keras.preprocessing.image.ImageDataGenerator(rescale=1./255.)
# Create a training image iterator that will be loaded in a small batch size. Resize all images to a #standard size.
val_it = datagen.flow_from_directory(val_img_dir, batch_si