动物分类识别教程+分类释义+界面展示-2

# nodelookup.py

import tensorflow.compat.v1 as tf
tf.disable_v2_behavior


class NodeLookup(object):
    def __init__(self):
        labels_path = 'inception_model/imagenet_2012_challenge_label_map_proto.pbtxt'
        uids_path = 'inception_model/imagenet_synset_to_human_label_map.txt'
        self.node_lookup = self.load(labels_path, uids_path)

    @staticmethod
    def _load(labels_path, uids_path):
        uid_to_human = {}
        for line in tf.gfile.GFile(uids_path).readlines():
            items = line.strip('\n').split('\t')
            uid_to_human[items[0]] = items[1]
        node_id_to_uid = {}
        for line in tf.gfile.GFile(labels_path).readlines():
            if line.startswith('  target_class:'):
                target_class = int(line.split(': ')[1])
            if line.startswith('  target_class_string:'):
                target_class_string = line.split(': ')[1]
                node_id_to_uid[target_class] = target_class_string[1:-2]
        node_id_to_name = {}
        for key, val in node_id_to_uid.items():
            name = uid_to_human[val]
            node_id_to_name[key] = name
        return node_id_to_name

    def id_to_string(self, node_id):
        if node_id not in self.node_lookup:
            return ''
        return self.node_lookup[node_id]

import tensorflow.compat.v1 as tf

tf.disable_v2_behavior
import numpy as np
import nodelookup


class TensorflowPredictor():
    def __init__(self):
        self.sess = tf.Session()
        with tf.gfile.FastGFile('./inception_model/classify_image_graph_def.pb', 'rb') as f:
            graph_def = tf.GraphDef()  # 定义一个计算图
            graph_def.ParseFromString(f.read())  #
            tf.import_graph_def(graph_def, name='')
        self.softmax_tensor = self.sess.graph.get_tensor_by_name('softmax:0')

    def predict_image(self, image_path):
        # 载入图片
        image_data = tf.gfile.FastGFile(image_path, 'rb').read()
        predictions = self.sess.run(self.softmax_tensor, {'DecodeJpeg/contents:0': image_data})  # 图片格式是jpg格式
        predictions = np.squeeze(predictions)  # 把结果转为1维
        # 打印图片路径及名称
        res_str = ''
        res_str += '图片路径: ' + image_path + '\n'
        # 排序
        top_k = predictions.argsort()[-5:][::-1]
        node_lookup = nodelookup.NodeLookup()
        for node_id in top_k:
            # 获取分类名称
            name_str = node_lookup.id_to_string(node_id)
            # 获取该分类的置信度
            score = predictions[node_id] * 100
            res_str += '(%.2f' % (score) + '%), ' + name_str + '\n'
        return res_str

import os
import tkinter
from tkinter import *
from tkinter import filedialog
from PIL import ImageTk
from translate import Translator

import get_Inception_model
from tensorflow_predictor import TensorflowPredictor

root = tkinter.Tk()  # 生成root主窗口
root.title("图像分类")  # 设置窗体标题
root.geometry("800x800")  # 设置窗体大小

if not os.path.exists('./inception_model/classify_image_graph_def.pb'):  # 如果没下载model，则下载model
    get_Inception_model.download_inception_model()  # 下载model

translator = Translator(to_lang="chinese")  # 新建Translator对象

def translator_prediction_result(pre_res):  # 翻译模块
    res = pre_res.split("\n")[0] + '\n'
    for line in pre_res.split("\n")[1:-1]:
        s = translator.translate(line.split(',')[1])
        res += line + " (机翻结果: " + s + ")\n"
    return res  # 返回翻译结果


img_label = Label(root, width='800', height='533')  # 这是是显示预测图片的全局变量
res_label = Label(root)  # 这是是显示预测文字的全局变量
pdt = TensorflowPredictor()  # 新建预测类(自己写的)


def selector_image():  # 选择图片按钮点击发生的事件
    img_path = filedialog.askopenfilename(initialdir='./images')  # 弹窗选择图像文件返回图像地址
    pre_res = pdt.predict_image(image_path=img_path)  # 利用地址调用预测函数返回结果字符串
    pre_res = translator_prediction_result(pre_res)  # 机器翻译结果字符串
    photo = ImageTk.PhotoImage(file=img_path)
    img_label.config(imag=photo)  # 更新图片
    img_label.pack()
    res_label.config(text=pre_res, justify=LEFT)  # 更新文字
    res_label.pack()
    root.mainloop()  # 进入消息循环
    return


btn_sel = tkinter.Button(root, text='选择图片', command=selector_image)  # 选择图片按钮
btn_sel.pack()

root.mainloop()  # 进入消息循环（必需组件）