app.py

import os
from flask import Flask, flash, request, redirect, url_for, render_template
from werkzeug.utils import secure_filename
import math

#  export FLASK_APP=app
#  flask run

arquivo_modelo = 'Model_2021_CNN_Xception-V09.hdf5' #'Model_2021_CNN_VGG19-V01.hdf5' # 'model_Titan-v02.hdf5' Só CCN


UPLOAD_FOLDER = '/tmp'
ALLOWED_EXTENSIONS = {'txt', 'pdf', 'png', 'jpg', 'jpeg', 'gif'}

def escolhe_lesao_aleatoria():
    import glob
    from random import seed
    from random import randint

    arquivos = list(glob.glob("static/tmp/*.*"))
    arquivos = [ arquivo.split('/')[2] for arquivo in arquivos]

    lesao = randint(0,len(arquivos)-1)
    print(lesao)
    return arquivos[lesao]


def prever_doencas_de_pele(model, file):
    import numpy as np
    from PIL import Image
    import pandas as pd

    folder = 'static/tmp/'

    dict_idx_doenca = {0: ['Actinic keratoses', 'Queratose Actínica'],
     1: ['Basal cell carcinoma', 'Carcinoma de Células Basais' ],
     2: ['Benign keratosis-like lesions ', 'Queratoses Benignas'],
     3: ['Dermatofibroma', 'Dermatofibroma'], # (Histiocitoma Fibroso Benigno)' ],
     4: ['Melanocytic nevi', 'Nevo Melanócito (Sinal)'], # (Nevo Pigmentado, Sinal)
     5: [ 'Melanoma', 'Melanoma'],
     6: ['Vascular lesions', 'Lesões de Pele Vasculares'],
     7: ['Acne', 'Acne'],
     8: ['AlopeciaAreata', 'AlopeciaAreata']}

    indices = []
    doencas_en = []
    doencas_pt = []
    for idx, doenca in (dict_idx_doenca.items()):
        indices.append(idx)
        doencas_en.append(doenca[0])
        doencas_pt.append(doenca[1])

    media_scale_image = 158.4125188825441
    std_scale_image = 47.42283803971779


    x = folder + file
    #x_pred = np.asarray(Image.open(x).resize((100,75)))

    SIZE = 299 # 224
    x_pred = np.asarray(Image.open(x).resize((SIZE,SIZE)))
    x_pred = x_pred.reshape(1, SIZE, SIZE, 3)
    x_pred = (x_pred - media_scale_image) / std_scale_image
    #classe = model.predict_classes(x_pred)[0]
    pred = np.argmax(model.predict(x_pred), axis=-1)
    probs = model.predict(x_pred)[0]

    probs = np.array(probs) * 100
    df = pd.DataFrame()
    df['probs'] = probs
    print('probs:', probs )
    df['probs'] = df['probs'].apply(lambda x : int(x))
    df['doenca_en'] = doencas_en
    df['doenca_pt'] = doencas_pt
    df['idx'] = indices
    df_ordenado = df.sort_values(by=['probs'], ascending=False).reset_index()
    df_ordenado = df_ordenado[ df_ordenado.probs > 0]

    numero_probilidades_maior_que_zero = len(df_ordenado)
    if numero_probilidades_maior_que_zero > 3:
        numero_probilidades_maior_que_zero = 3
    probs = df_ordenado['probs'][:numero_probilidades_maior_que_zero]
    doencas = df_ordenado['doenca_pt'][:numero_probilidades_maior_que_zero]

    #probs = df_ordenado['probs'][:3]
    #doencas = df_ordenado['doenca_pt'][:3]

    #print('diagnóstico:', doenca, ' - prob:', prob)
    #print(doenca)
    #print(prob)

    return doencas, probs


def allowed_file(filename):
    return '.' in filename and \
           filename.rsplit('.', 1)[1].lower() in ALLOWED_EXTENSIONS


app = Flask(__name__, template_folder='templates')
app.secret_key = "super secret key"

app.config['UPLOAD_FOLDER'] = 'static/tmp'
app.config['MAX_CONTENT_LENGTH'] = 16 * 1024 * 1024


probs = []
classesprev = []
model = None

app.add_url_rule('/static', view_func=app.send_static_file)
@app.route('/', methods=['GET', 'POST'])
def upload_file():
    #from app import model
    global model
    import numpy as np
    import tensorflow as tf

    #from keras.models import load_model


    if model is None:
        print('carregando o modelo...')
        file_model = arquivo_modelo

        from tensorflow import keras
        # model = keras.models.load_model(file_model)
        model = tf.keras.models.load_model(file_model,
                custom_objects={'Functional':tf.keras.models.Model})

    

        #model = tf.keras.models.load_model(file_model)
        #model = load_model(file_model)
        print('modelo carregado.')


    UPLOAD_FOLDER = '/tmp'
    ALLOWED_EXTENSIONS = {'txt', 'pdf', 'png', 'jpg', 'jpeg', 'gif'}

    if request.method == 'POST':
        print('request == POST')
        d = request.form.to_dict(flat=False)
        print(d)

        if "photo" in d.keys() and  "prever_lesao" in d.keys() and d['photo'][0] != '': # request.form["prever_lesao"]:
            file  = request.form['photo']
            doencas, probs = prever_doencas_de_pele(model, file)

            return render_template("index.html",  file='tmp/'+file,  probs=probs, classesprev=doencas)

        else:
            file = escolhe_lesao_aleatoria()
            print(file)
            doencas, probs = prever_doencas_de_pele(model, file)

            return render_template("index.html",  file='tmp/'+file,  probs=probs, classesprev=doencas)


    else:
        print("elsseeeeee")
        file = escolhe_lesao_aleatoria()
        doencas, probs = prever_doencas_de_pele(model, file)

        return render_template("index.html",  file='tmp/'+file,  probs=probs, classesprev=doencas) #, upload_file=global_file)

@app.route('/about/')
def about():
    return render_template('About.html')

if __name__ == "__main__":

    app.config['SESSION_TYPE'] = 'filesystem'
    port = int(os.environ.get("PORT", 5000))
    app.debug = True
    app.run(host='0.0.0.0', port=port)