0

I tried to do human pose action recognition model, I referred this model

I like to use LSTM for this model. So I have made some changes in train.py

My train.py code:

import pandas as pd
from enum import Enum
import numpy as np
from sklearn.preprocessing import LabelEncoder
from sklearn.model_selection import train_test_split
from keras.utils import np_utils
from keras.models import Sequential
from keras.layers import Dense, Dropout
from keras.layers.normalization import BatchNormalization
from keras.optimizers import Adam
from keras.models import load_model
from keras.layers import LSTM

class Actions(Enum):

    sit = 0
    stand = 1
    walk = 2
    sleep= 3

raw_data = pd.read_csv('7537real1.csv', header=0)
dataset = raw_data.values

X = dataset[0:7537, 0:36].astype(float)  
Y = dataset[0:7537, 36]

encoder_Y = [0]* 4479 + [1]* 1425 + [2] * 1164 + [3] * 468
dummy_Y = np_utils.to_categorical(encoder_Y)
X_train, X_test, Y_train, Y_test = train_test_split(X, dummy_Y, test_size=0.1, random_state=9)

model = Sequential()
model.add(LSTM(4, input_shape=(36,1)))
model.add(Dense(units=4, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer=Adam(0.0001), metrics=['accuracy'])
model.fit(X_train, Y_train, batch_size=32, epochs=500, verbose=1, validation_data=(X_test, Y_test))
model.save('7537real1.h5')

My data set has 36 features and class attribute (labels:0,1,2,3) And totally there are 7537 records in the dataset. When I tried to build the LSTM sequential classification model I got value error.

Also I have attached dataset sample as screenshot (csv file). enter image description here

How to reshape the data(array) set for this model and how to build the LSTM sequential model?

Malathi K
  • 71
  • 12
  • is your dataset sequential? and if so what is the length of the individual sequence? – Mitiku Mar 22 '19 at 06:13
  • @Mitiku Please check above, now i have attached screenshot of my dataset. class label "0" has 4478 records, class label "1" has 1425 records, class label "2" has 1164 records, class label "3" has 469 records. – Malathi K Mar 22 '19 at 08:55

0 Answers0