gradio_app.py

import gradio as gr
import openai
import numpy as np
import tensorflow as tf
import keras
from PIL import Image
import requests
import json
from json import JSONEncoder
from datetime import datetime


myControls = {
    "ResultControl":None,
    "Feedback":None,
    "AdditionalInfo":None
}


dataToSend = {
    "FileContent":None,
    "PlantName":None,
    "Comments":None
}

imageData = []

class NumpyEncoder(JSONEncoder):
    def default(self, obj):
        if isinstance(obj, np.ndarray) :
            return obj.tolist()
        return JSONEncoder.default(self,obj)
    


def uploadFile() :
    global dataToSend

    npArray = np.asarray(imageData)
    fileContent = json.dumps(npArray, cls=NumpyEncoder)
    dataToSend["FileContent"] = fileContent

    payload = json.dumps(dataToSend)


    r = requests.post("http://127.0.0.1:5000/todb", data=payload)
    return r

def saveStats(predictionStatus) :
    d = {
        'Time': str(datetime.now()),
        'PredictionStatus':None
    }

    if predictionStatus == 'Satisfied' :
        d['PredictionStatus'] = 1
    else :
        d['PredictionStatus'] = 0

    r = requests.post("http://127.0.0.1:5000/predictionstats", data=json.dumps(d))
    return r

    
def predict(imageToProcess):
    global imageData
    reply = "Nothing to display"
    try :
    
        openai.api_key = "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"
        message = "Mango Plant Diseases" 
        if message: 
            messages = []
            messages.append( 
                {"role": "user", "content": message}, 
            ) 
            chat = openai.ChatCompletion.create( 
                model="gpt-3.5-turbo", messages=messages 
            ) 
        
        reply = chat.choices[0].message.content 
    except :
        pass

    imageData = imageToProcess
    print("Image Dimensions", imageData.height, imageData.width)

    return ["No Disease", reply]

def submitFeedback(correctOrWrong, plantName, userData):
    global dataToSend

    print(correctOrWrong)


    if correctOrWrong == "Not Satisfied" :
        dataToSend["PlantName"] = plantName
        dataToSend["Comments"] = userData
        dataToSend["FileContent"] = json.dumps(np.asarray(imageData).tolist())
        r = uploadFile()

        if r != None :
            res = json.loads(r.text)
            gr.Warning("Data Submitted for learning :" + res["Status"])
        else :
            gr.Error("Failed to upload the file for learning")

    saveStats(correctOrWrong)

with gr.Blocks(allow_flagging="manual") as app :

    gr.Markdown(
    """
        # AI based plant Disease Detection Application
       
    """
    )
    myControls["ImageInput"] = gr.Image(type="pil")

    controls = []

    myControls["ResultControl"] = gr.Textbox(label='Possible Disease could be ')
    myControls["AdditionalInfo"] = gr.TextArea(label='Additional Info')
    controls.append(myControls["ResultControl"])
    controls.append(myControls["AdditionalInfo"])
    

    predictBtn = gr.Button(value='Predict')
    predictBtn.click(predict, inputs=[myControls["ImageInput"]], outputs=controls)


    gr.Markdown()

    myControls["PredictionSelection"] = gr.Radio(["Satisfied", "Not Satisfied"], label="Feedback", info="Are you satisfied with the prediction?")
    #myControls["Feedback"] = gr.Checkbox(label="Is prediction wrong? If so, please provide the proper classification")
    myControls["PlantName"] = gr.Textbox(label='Specify the name of the plant')
    myControls["UserInput"] = gr.Textbox(label='What is the correct classification?')
    feedbackBtn = gr.Button(value='Submit Feedback')
    feedbackBtn.click(submitFeedback, inputs =[myControls["PredictionSelection"], myControls["PlantName"], myControls["UserInput"]])


    




    app.queue().launch()