不支持在图形模式下调用“myModel.predict”

Question

Had some issues on this when developing image recognition website

[2021-07-19 02:27:21,880] ERROR in app: Exception on /uploads/comedy3.jpg [GET]
Traceback (most recent call last):
  File "C:\Users\Ram Raj\anaconda3\envs\Tutorial\lib\site-packages\flask\app.py", line 2447, in wsgi_app
    response = self.full_dispatch_request()
  File "C:\Users\Ram Raj\anaconda3\envs\Tutorial\lib\site-packages\flask\app.py", line 1952, in full_dispatch_request
    rv = self.handle_user_exception(e)
  File "C:\Users\Ram Raj\anaconda3\envs\Tutorial\lib\site-packages\flask\app.py", line 1821, in handle_user_exception
    reraise(exc_type, exc_value, tb)
  File "C:\Users\Ram Raj\anaconda3\envs\Tutorial\lib\site-packages\flask\_compat.py", line 39, in reraise
    raise value
  File "C:\Users\Ram Raj\anaconda3\envs\Tutorial\lib\site-packages\flask\app.py", line 1950, in full_dispatch_request
    rv = self.dispatch_request()
  File "C:\Users\Ram Raj\anaconda3\envs\Tutorial\lib\site-packages\flask\app.py", line 1936, in dispatch_request
    return self.view_functions[rule.endpoint](**req.view_args)
  File "C:\Users\Ram Raj\Desktop\Ai project\website.py", line 83, in uploaded_file
    result = myModel.predict(test_image)
  File "C:\Users\Ram Raj\anaconda3\envs\Tutorial\lib\site-packages\tensorflow\python\keras\engine\training.py", line 1671, in predict
    version_utils.disallow_legacy_graph('Model', 'predict')
  File "C:\Users\Ram Raj\anaconda3\envs\Tutorial\lib\site-packages\tensorflow\python\keras\utils\version_utils.py", line 130, in disallow_legacy_graph
    raise ValueError(error_msg)
ValueError: Calling `Model.predict` in graph mode is not supported when the `Model` instance was constructed with eager mode enabled. Please construct your `Model` instance in graph mode or call `Model.predict` with eager mode enabled.

Here is the code

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

import numpy as np

from tensorflow.keras.preprocessing import image
from tensorflow.keras.models import load_model
from keras.backend import set_session
import tensorflow as tf

app = Flask(__name__)

def load_model_from_file():
    mySession = tf.compat.v1.Session()
    set_session(mySession)
    myModel = load_model('saved_model.h5')
    myGraph = tf.compat.v1.get_default_graph()
    return (mySession, myModel, myGraph)

I have just showed part where it shows wrong only. Hope able to understand it

@app.route('/uploads/<filename>')
def uploaded_file(filename):
    test_image = image.load_img(UPLOAD_FOLDER+"/"+filename, target_size =(150,150))
    test_image = image.img_to_array(test_image)
    test_image = np.expand_dims(test_image, axis=0)

    mySession = app.config['SESSION']
    myModel = app.config['MODEL']
    myGraph = app.config['GRAPH']
    with myGraph.as_default():
        set_session(mySession)
        result = myModel.predict(test_image)
        image_src = "/"+UPLOAD_FOLDER +"/"+filename
        if result[0] < 0.5 :
            answer = "<div class='col text-center'><img width='150' height='150' src='"+image_src+"' class='img-thumbnail' /><h4>guess:"+x+" "+str(result[0])+"</h4></div><div class='col'></div><div class='w-100'></div>"     
        else:
            answer = "<div class='col'></div><div class='col text-center'><img width='150' height='150' src='"+image_src+"' class='img-thumbnail' /><h4>guess:"+y+" "+str(result[0])+"</h4></div><div class='w-100'></div>"     
        results.append(answer)
        return render_template('index.html',myX=x,myY=y,mySampleX=samplex,mySampleY=sampley,len=len(results),results=results)



def main():
    (mySession, myModel, myGraph) = load_model_from_file()
    
    app.config['SECRET_KEY'] = 'super secret key'
    app.config['SESSION'] = mySession
    app.config['MODEL'] = myModel
    app.config['GRAPH'] = myGraph
    
    app.config['UPLOAD_FOLDER'] = UPLOAD_FOLDER
    app.config['MAX_CONTENT_LENGTH'] = 16 * 1024 * 1024
    app.run()


results = []

main()

Any idea on where is the fault that i did ? As i couldn't understand much when i searched it at tensorflow website as well. The version that i'm using was 2.5.0 thank you in advance for the help

Answer 1

ValueError: Calling Model.predict in graph mode is not supported when the Model instance was constructed with eager mode enabled. Please construct your Model instance in graph mode or call Model.predict with eager mode enabled.

正如error所说，加载模型图模式并调用model.predict函数

将模型加载到图表中，如下所示：

with graph.as_default():
    #Load your model
    #call model.predict
    Model.predict()
    return