'SGD' object has no attribute 'get_updates'

[beingSherlock]

AttributeError Traceback (most recent call last)
in
1 # train weights (output layers or 'heads')
----> 2 model.train(dataset_train, dataset_train, learning_rate=0.001, epochs=5, layers='heads')

~\Downloads\Mask-R-CNN-using-Tensorflow2-main\mrcnn\model.py in train(self, train_dataset, val_dataset, learning_rate, epochs, layers, augmentation, custom_callbacks, no_augmentation_sources)
2367 workers = multiprocessing.cpu_count()
2368
-> 2369 self.keras_model.fit(
2370 train_generator,
2371 initial_epoch=self.epoch,

~\AppData\Roaming\Python\Python38\site-packages\keras\engine\training_v1.py in fit(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_freq, max_queue_size, workers, use_multiprocessing, **kwargs)
854
855 func = self._select_training_loop(x)
--> 856 return func.fit(
857 self,
858 x=x,

~\AppData\Roaming\Python\Python38\site-packages\keras\engine\training_generator_v1.py in fit(self, model, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_freq, max_queue_size, workers, use_multiprocessing)
645 y, sample_weight, validation_split=validation_split
646 )
--> 647 return fit_generator(
648 model,
649 x,

~\AppData\Roaming\Python\Python38\site-packages\keras\engine\training_generator_v1.py in model_iteration(model, data, steps_per_epoch, epochs, verbose, callbacks, validation_data, validation_steps, validation_freq, class_weight, max_queue_size, workers, use_multiprocessing, shuffle, initial_epoch, mode, batch_size, steps_name, **kwargs)
280
281 is_deferred = not model._is_compiled
--> 282 batch_outs = batch_function(*batch_data)
283 if not isinstance(batch_outs, list):
284 batch_outs = [batch_outs]

~\AppData\Roaming\Python\Python38\site-packages\keras\engine\training_v1.py in train_on_batch(self, x, y, sample_weight, class_weight, reset_metrics)
1178
1179 self._update_sample_weight_modes(sample_weights=sample_weights)
-> 1180 self._make_train_function()
1181 outputs = self.train_function(ins)
1182

~\AppData\Roaming\Python\Python38\site-packages\keras\engine\training_v1.py in _make_train_function(self)
2282 with backend.name_scope("training"):
2283 # Training updates
-> 2284 updates = self.optimizer.get_updates(
2285 params=self._collected_trainable_weights,
2286 loss=self.total_loss,

AttributeError: 'SGD' object has no attribute 'get_updates'

I'm new to Machine Learning please help me with this error.

[darwinOne]

@beingSherlock You must upgrade Tensorflow to version 2.5x or above. Its work for me.

[beingSherlock]

Ok thanks for the help. Had to change some things on system files, it worked fine after that.

[RomanProkazov]

@beingSherlock Which things did you change on system files? Upgrading to tensorflow 2.5 is not a solution for me, because it is gonna be incompatable with other packages in my environment. I use tensorflow 2.12.0, facing the same issue.

[ShSi1001]

@beingSherlock can you tell what exactly did you change?

[akshayr5705]

I had the same error .
i've done some changes in the model.py file inside the mrcnn folder worked for me.

specifically i changed the line

2168-> optimizer = tf.keras.optimizers.SGD(....)
to

2168-> optimizer = tf.keras.optimizers.legacy.SGD(...)

[arabinda90]

Ok thanks for the help. Had to change some things on system files, it worked fine after that.

@beingSherlock Can you please tell me what actually needs to change?

[amitkm10]

Can you please tell me what exactly need to do in the code or configuration

[kanzajamil]

@beingSherlock how you resolved it? I am getting the same error and tf.keras.optimizers.legacy.SGD is not working for me! also I'm using version 2.14.0 . Please help anyone who resolved it using version 2.14.0.