Thanks for sharing your work. That is really helpful. However in the tutorial "Intent Recognition with BERT using Keras and TensorFlow 2", why the valid loss/acc do not change by epoches? The training loss/acc are improved but the model seems to have good performance on the valid dataset before any finetuning.

Hi, sorry for a basic question, i think, but, i see no material or explanation of the 'step' variable. Just trying to edit for my code but unsure exactly. I understand time_step = how many steps does my time series inputs. But steps?

Thanks,
Tom

Hi. I am using your google colab to test object detection on my own dataset which I have uploaded on your google colab folder. Everything runs fine before the training step i.e. before "Downloaded pretrained model to ./snapshots/_pretrained_model.h5" step.

But when I am running the command "!keras_retinanet/bin/train.py --freeze-backbone --random-transform --weights {PRETRAINED_MODEL} --batch-size 8 --steps 500 --epochs 10 csv annotations.csv classes.csv" on the colab, the model is running only for one epoch and only one file "resnet50_csv_01.h5" is getting generated and the iterations is not running after 1st epoch.

Even I tried the same in my local machine and the same issue is happening i.e. after 1st epoch the process is stopping even though number of epochs in the command is 10. In my local machine and even in the google colab I have seen that tensorflow version is 2.4.

Can you let me know what is the issue.

I want to ask for your advice. I want to identify the bottles in the trash can. The real image is about this:

The most complex part of this scene is the dense shielding and stacking of bottles. Our purpose is to tell the customer who threw the bottle what kind of bottle he threw in, and it's better to return the location information of the bottle at the bottom of the trash can. At present, I use Yolo series for testing, and the effect is always unsatisfactory.

I want to know does you have any suggestions to improve the detection and classification ability in this scenario?

Much obliged for your assistance.

In the time series, i dont see Timestamp being used. Is that intentional ?

how can i convert .h5 file to .pb,
i want to use trained model in java application,

Any advise or guidance would be greatly appreciated..!!
Thanks

I was using the file "13.time-series-human_activity_recognition.ipynb"

And trying to convert the model to TfLite so that i can use it in Android Project am getting an exception.

Using Below code.

converter = tf.lite.TFLiteConverter.from_keras_model(model)
tflite_model = converter.convert()

# Save the model.
with open('model.tflite', 'wb') as f:
  f.write(tflite_model)

If you can suggest what could be the issue.

Observe the same issue for

model.save("./")

converter = tf.lite.TFLiteConverter.from_saved_model("./")
tflite_model = converter.convert()

converter.convert() is the line which throws

!keras_retinanet/bin/train.py --freeze-backbone --random-transform --weights {PRETRAINED_MODEL} --batch-size 8 --steps 500 --epochs 10 csv annotations.csv classes.csv

Traceback (most recent call last):
File "keras_retinanet/bin/train.py", line 540, in
main()
File "keras_retinanet/bin/train.py", line 500, in main
config=args.config
File "keras_retinanet/bin/train.py", line 114, in create_models
model = model_with_weights(backbone_retinanet(num_classes, num_anchors=num_anchors, modifier=modifier), weights=weights, skip_mismatch=True)
File "keras_retinanet/bin/../../keras_retinanet/models/resnet.py", line 38, in retinanet
return resnet_retinanet(*args, backbone=self.backbone, **kwargs)
File "keras_retinanet/bin/../../keras_retinanet/models/resnet.py", line 99, in resnet_retinanet
resnet = keras_resnet.models.ResNet50(inputs, include_top=False, freeze_bn=True)
File "/usr/local/lib/python3.6/dist-packages/keras_resnet/models/_2d.py", line 188, in ResNet50
return ResNet(inputs, blocks, numerical_names=numerical_names, block=keras_resnet.blocks.bottleneck_2d, include_top=include_top, classes=classes, *args, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/keras_resnet/models/_2d.py", line 66, in ResNet
x = keras_resnet.layers.BatchNormalization(axis=axis, epsilon=1e-5, freeze=freeze_bn, name="bn_conv1")(x)
File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/base_layer.py", line 922, in call
outputs = call_fn(cast_inputs, *args, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/autograph/impl/api.py", line 265, in wrapper
raise e.ag_error_metadata.to_exception(e)
TypeError: in user code:

/usr/local/lib/python3.6/dist-packages/keras_resnet/layers/_batch_normalization.py:17 call  *
    return super(BatchNormalization, self).call(training=(not self.freeze), *args, **kwargs)

TypeError: type object got multiple values for keyword argument 'training'

I am getting
NotImplementedError: Cannot convert a symbolic Tensor (bidirectional/forward_cu_dnnlstm/strided_slice:0) to a numpy array. This error may indicate that you're trying to pass a Tensor to a NumPy call, which is not supported

for
model.add(Bidirectional(CuDNNLSTM(WINDOW_SIZE, return_sequences=True),
input_shape=(WINDOW_SIZE, X_train.shape[-1])))

I am running Tensorflow 2.4.0rc0 and numpy 1.21.4

i saw somewhere the moving back to numpy==1.19.5 could resolve it, but it did not have an impact

Can you advise the best direction here?

model = create_model(data.max_seq_len, bert_ckpt_file) gives error

TypeError Traceback (most recent call last)
in ()
----> 1 model = create_model(data.max_seq_len, bert_ckpt_file)

5 frames
/usr/local/lib/python3.7/dist-packages/tensorflow/python/autograph/impl/api.py in wrapper(*args, **kwargs)
693 except Exception as e: # pylint:disable=broad-except
694 if hasattr(e, 'ag_error_metadata'):
--> 695 raise e.ag_error_metadata.to_exception(e)
696 else:
697 raise

TypeError: in user code:

/usr/local/lib/python3.7/dist-packages/bert/model.py:80 call  *
    output           = self.encoders_layer(embedding_output, mask=mask, training=training)
/usr/local/lib/python3.7/dist-packages/keras/engine/base_layer.py:1030 __call__  **
    self._maybe_build(inputs)
/usr/local/lib/python3.7/dist-packages/keras/engine/base_layer.py:2659 _maybe_build
    self.build(input_shapes)  # pylint:disable=not-callable
/usr/local/lib/python3.7/dist-packages/bert/transformer.py:209 build
    self.input_spec = keras.layers.InputSpec(shape=input_shape)
/usr/local/lib/python3.7/dist-packages/keras/engine/base_layer.py:2777 __setattr__
    super(tf.__internal__.tracking.AutoTrackable, self).__setattr__(name, value)  # pylint: disable=bad-super-call
/usr/local/lib/python3.7/dist-packages/tensorflow/python/training/tracking/base.py:530 _method_wrapper
    result = method(self, *args, **kwargs)
/usr/local/lib/python3.7/dist-packages/keras/engine/base_layer.py:1297 input_spec
    'Got: {}'.format(v))

TypeError: Layer input_spec must be an instance of InputSpec. Got: InputSpec(shape=(None, 38, 768), ndim=3)