基于flask框架，完成Unet图像分割及EfficientNet图像分类工作（python）

from flask import Flask, make_response, render_template, request, jsonify from static.model_fenlei.model import efficientnet_b0 as create_model import torch from torchvision import transforms import json from PIL import Image from static.model_fenge.unet import Unet import datetime import os import numpy as np import matplotlib.pyplot as plt import openpyxl from flask import url_for # EfficientNet模型代码植入 device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") save_time = datetime.datetime.now().strftime('%Y%m%d%H%M%S%f') img_size = {"B0": 224, "B1": 240, "B2": 260, "B3": 300, "B4": 380, "B5": 456, "B6": 528, "B7": 600} num_model = "B0" json_path = './static/model_fenlei/class_indices_butterfly.json' with open(json_path, "r") as f: class_indict = json.load(f) model = create_model(num_classes=100).to(device) model_weight_path = "./static/model_fenlei/best_model_loss_butterfly.pth" model.load_state_dict(torch.load(model_weight_path, map_location=device)) model.eval() data_transform = transforms.Compose( [transforms.Resize(img_size[num_model]), transforms.CenterCrop(img_size[num_model]), transforms.ToTensor(), transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])]) # U-Net模型代码植入 unet = Unet() count = False name_classes = ["background", "蝴蝶"] app = Flask(__name__) @app.route('/', methods=['GET']) # 装饰器，修饰我们底下的函数，让我们的函数具备下面的功能 # 程序的路由，当浏览器访问我们的根路径的时候，会执行下面的函数 def main_page(): # put application's code here res = make_response(render_template('log_in.html')) return res @app.route('/login', methods=['POST', 'GET']) def login(): if request.method == 'POST': name = request.form["user"] password = request.form["password"] if name == 'SUJUNYI' and password == '12345': return render_template("classifier.html") # 核实密码以及用户名 else: return render_template("404.html") else: return render_template("404.html") # 2023年9月2号添加，为了显示预测最大值的病媒生物的信息 def find_info_by_name(excel_file, name_to_find): try: # 打开Excel文件 workbook = openpyxl.load_workbook(excel_file) sheet = workbook.active # 初始化一个列表来存储匹配的信息 info_list = [] # 遍历每一行，查找匹配的名称并将信息添加到列表中 for row in sheet.iter_rows(values_only=True): if row[0] == name_to_find: info = [row[0], row[1]] info_list.append(info) # 关闭Excel文件 workbook.close() return info_list except Exception as e: print("发生错误:", e) return [] # 替换成你的Excel文件路径 excel_file = "E:\Web开发\病媒识别4\病媒生物介绍.xlsx" @app.route('/predict', methods=['GET', 'POST']) # 后端图片上传 def predict(): ########################预测多张，并且把每一张的概率都给打印在网页上############################### if request.method == 'GET': res = make_response(render_template('classifier.html')) return res elif request.method == 'POST': if 'myImg' in request.files: uploaded_files = request.files.getlist('myImg') # Get list of uploaded files results_list = [] # List to store results for each image results_list2 = [] integrated_results = {} segmented_image_urls = [] for objFile in uploaded_files: strFileName = objFile.filename strFilePath = "./static/myimages/" + save_time + strFileName objFile.save(strFilePath) strFilePath2 = "./static/myimages_fengehou/" + save_time + strFileName segmented_image_url = url_for('static', filename='myimages_fengehou/' + save_time + strFileName) segmented_image_urls.append(segmented_image_url) # 上面这部分属于对于传入的图片的保存 # 下面这部分属于对于保存的图片进行图像分割，再将其传入到分类模型得出结果 img = Image.open(strFilePath) r_image = unet.detect_image(img, count=count, name_classes=name_classes) r_image.save(strFilePath2) img = Image.open(strFilePath2) img = data_transform(img) img = torch.unsqueeze(img, dim=0) with torch.no_grad(): output = torch.squeeze(model(img.to(device))).cpu() predict = torch.softmax(output, dim=0) print(predict.shape) predict_topk_values, predict_topk_indices = torch.topk(predict, k=5, dim=0) predict_topk_indices = predict_topk_indices.cpu().numpy() # Convert tensor indices to numpy array predict_class_names = [class_indict[str(idx.item())] for idx in predict_topk_indices] predict_probabilities = predict_topk_values.numpy() # results系列的代码，把每一个图片的前五种都显示出来，然后拼到results_list当中 # results2系列的代码，把每一张图片的前五种概率进行整合，然后拼接得出所有的平均概率，然后拼到results_list2当中 results = [] results2 = [] for class_name, probability in zip(predict_class_names, predict_probabilities): percent_probability = probability * 100 result = " {}的概率大小为: {:.1f}%".format(class_name, percent_probability) results.append(result) result2 = (class_name, percent_probability) results2.append(result2) results_list.append(results) # Store results for current image results_list2.append(results2) # 遍历 results_list2 中的每个图片的结果列表 for image_results in results_list2: # 创建一个临时字典来存储当前图片的结果 temp_results = {} # 遍历当前图片的结果元组 for class_name, probability in image_results: if class_name not in temp_results: temp_results[class_name] = probability else: temp_results[class_name] += probability # 将当前图片的结果添加到整合的结果中，就是把每张图片的类别概率进行相加，如果没有这个类别就添加这个类别 for class_name, probability in temp_results.items(): if class_name not in integrated_results: integrated_results[class_name] = probability else: integrated_results[class_name] += probability # print(integrated_results) # 将字典中的值转换为列表 values = np.array(list(integrated_results.values())) # print(values) # 各个值除以，总的值 normalized_values = values / sum(values) * 100 # print(normalized_values) # 创建一个新的字典，包含归一化后的结果，就是加起来是100 normalized_results = {class_name: probability for class_name, probability in zip(integrated_resul