[FIXED] How to update weights of a base Keras PointNet model?

Issue

I trained a PointNet regression model and then saved it's weights by

model.save_weights('/home/rev9ai/aleef/Dainsta/model/beta_v0.weights.ckpt')

The saved files were

• beta_v0.weights.ckpt.data-00000-of-00001
• beta_v0.weights.ckpt.index

Now I want to update a base PointNet model using these weights Here is how the model was built using Tensorflow:

# Define the PointNet architecture for regression
def build_pointnet_regression_model(num_points=2048):
    inputs = keras.Input(shape=(num_points, 3))
    
    x = tnet(inputs, 3)
    x = conv_bn(x, 32)
    x = conv_bn(x, 32)
    x = tnet(x, 32)
    x = conv_bn(x, 32)
    x = conv_bn(x, 64)
    x = conv_bn(x, 64) #additional
    x = conv_bn(x, 64) #additional
    x = conv_bn(x, 512)
    x = layers.GlobalMaxPooling1D()(x)
    x = dense_bn(x, 256)
    # x = layers.Dropout(0.3)(x)
    # x = dense_bn(x, 128)
    # x = layers.Dropout(0.3)(x)

    # Regression output layer
    outputs = layers.Dense(1, activation="relu")(x)
    
    model = keras.Model(inputs=inputs, outputs=outputs, name="pointnet_regression")
    return model

# Create the PointNet regression model
model = build_pointnet_regression_model(NUM_POINTS)

# Compile the model for regression
model.compile(
    loss="mean_absolute_error", 
    optimizer=keras.optimizers.RMSprop(learning_rate=0.1),
    metrics=["mean_absolute_error"],
)

Solution

model.load_weights('location/to/weights.ckpt')

That's how you load weights. Just build/ define the model as usual and then load pre-saved weights.

Answered By - Aleef