Closed johnfoxy closed 2 years ago
I am trying to train my Unet Model, but I keep ending up with this error (below). Other losses work but not Jaccard, any help please?
"ValueError: Unknown loss function: bce_jaccard_loss. Please ensure this object is passed to the custom_objects argument. See https://www.tensorflow.org/guide/keras/save_and_serialize#registering_the_custom_object for details."
custom_objects
My code snippet failing...
`model = sm.Unet(BACKBONE, encoder_weights='imagenet') model.compile(optimizer='Adam', loss='bce_jaccard_loss', metrics=['iou_score'])
print(model.summary())
history=model.fit(x_train[0:500], y_train[0:500], batch_size=2, epochs=1000, verbose=1, validation_data=(x_val, y_val))`
I am trying to train my Unet Model, but I keep ending up with this error (below). Other losses work but not Jaccard, any help please?
"ValueError: Unknown loss function: bce_jaccard_loss. Please ensure this object is passed to the
custom_objectsargument. See https://www.tensorflow.org/guide/keras/save_and_serialize#registering_the_custom_object for details."My code snippet failing...
`model = sm.Unet(BACKBONE, encoder_weights='imagenet') model.compile(optimizer='Adam', loss='bce_jaccard_loss', metrics=['iou_score'])
print(model.summary())
history=model.fit(x_train[0:500], y_train[0:500], batch_size=2, epochs=1000, verbose=1, validation_data=(x_val, y_val))`