H! I am working on predicting two variables using time series forecasting autoencoder model. The dataset present the coordinates bounding box in video like x and y. So let's say we have 6 people in video, so that mean we will have 12 variable as input in the model for each image. The LSTM model doing a good job predicting just one variable. However, I want to predict two variables ( x and y) using autoencoder model. I am using karas to this work. This is the model that I use.
model = Sequential()
model.add(LSTM(64, activation='relu', input_shape=(trainX.shape[1], trainX.shape[2]), return_sequences=True))
model.add(LSTM(32, activation='relu', return_sequences=False))
model.add(Dropout(0.2))
model.add(Dense(trainY.shape[2]))
cp1 = ModelCheckpoint('model1/', save_best_only=True)
model.compile(loss='mse', optimizer=Adam(learning_rate=0.0001), metrics=['mae', 'mape'])
model.summary()
I want to predict two variables ( x and y) using autoencoder model. One of the suggestion was use two dimensional MSE, which is involved x and y values. But not sure how to do that. I will really appropriate any suggestion or command that can be helpful
