jschw

import sys
import json
import openai
from langchain.text_splitter import Document


# ===== Functions ======

def build_prompt(input_docs, user_input):
        

jschw

from tensorflow import keras
import tensorflow.keras.backend as K
from tensorflow.keras.models import Model
from tensorflow.keras.layers import Input, Dense

import numpy as np
import cv2
import matplotlib
import matplotlib.pyplot as plt

jschw

import onnxruntime as onnxrun

# Processing with ONNX Runtime
REAL_Prediction = np.array(sess.run(None, {WORD_InputName: REAL_InputVector}))

REAL_PredictionConfidence = REAL_Prediction[0, 0, np.argmax(REAL_Prediction)]
REAL_PredictionConfidence = round(REAL_PredictionConfidence*100, 2)

DINT_ResultVector = np.argmax(REAL_Prediction)

jschw

from tensorflow import keras
import onnxmltools

# Load and convert model to ONNX
model = keras.models.load_model('model_curvetypes.h5')

onnx_model = onnxmltools.convert_keras(model, model.name)
onnxmltools.utils.save_model(onnx_model, 'model_curvetypes.onnx')

jschw

PROGRAM PLC_PRG

VAR
	
	BOOL_O_JetsonTrigger : BOOL := FALSE;
	
	DINT_Step : DINT := 0;
	
	DINT_InputVectorLength : DINT := 20;
	

jschw

start_preprocessing = time.time()
prediction = model.predict(predict_arr)
end_preprocessing = time.time()

pred_conf = prediction[0, np.argmax(prediction)]
pred_conf = round(pred_conf*100, 2)

pred_time = round((end_preprocessing-start_preprocessing), 3)

print('Predicted class no.: ' + str(np.argmax(prediction)))

jschw

# Specific sample
#predict_arr = X_test[15].reshape(1,20)
#predict_ground_truth = y_onehot_test[15]

# Random sample
rand_index = np.random.choice(len(X_test), 1)
predict_arr = X_test[rand_index].reshape(1,20)
predict_ground_truth = y_onehot_test[rand_index]

jschw

def get_model():
    model = Sequential()
    model.add(Dense(64, input_dim=20, activation='relu'))
    model.add(Dropout(0.2))
    model.add(Dense(128, activation='relu'))
    model.add(Dropout(0.3))
    model.add(Dense(32, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(16, activation='relu'))
    model.add(Dropout(0.5))

jschw

# Split x/y
X = data.values[:,0:-1]
Y = data.values[:,-1]

X = X.astype('float32')

# Encode label vector
encoder = LabelEncoder()
encoded_Y = encoder.fit_transform(Y)

jschw

data = pd.read_csv('curve_dataset_train.csv', index_col=False)
data = data.sample(frac=1, random_state=42).reset_index()
data.drop(['index'], 1, inplace=True)
print(data.shape)
data[:9][['shape', 'x1', 'y1','x2', 'y2','x3', 'y3','x4', 'y4','x5', 'y5','x6', 'y6','x7', 'y7','x8', 'y8','x9', 'y9','x10', 'y10']]