CTC or “connectionist temporal classification” is a machine learning technique for mapping dense input data to shorter output sequences in the same order.
Questions tagged [ctc]
85 questions
0
votes
1 answer
CTC_Loss for Model training in Tensorflow2.0
I am working on a project where in we have to use ctc_batch_cost for Loss. I defined the function that returns the CTC loss and tried to use it in 'model.compile' statement. But here, I could not figure out how to get 'y_pred'. Can someone help me…
Raja Malyala
- 89
- 2
- 6
0
votes
1 answer
Loading custom CTC layer from h5 file in Keras
I have a CTCLayer class like this:
class CTCLayer(layers.Layer):
def __init__(self, name=None):
super().__init__(name=name)
self.loss_fn = keras.backend.ctc_batch_cost
def call(self, y_true, y_pred):
# Compute the training-time loss…
jayiswhuuu
- 41
- 3
0
votes
0 answers
RuntimeError: stack expects each tensor to be equal size, but got [7] at entry 0 and [6] at entry 2
Trying to train a model with custom images to recognize license plate characters, but this is the error:
model = model.train()
loss_history = []
for epoch in range(10):
pbar = tqdm(dataloader_train)
for batch_image, batch_carnumber in…
Sebastián
- 437
- 5
- 19
0
votes
1 answer
CNN with CTC loss
I want to extract features using a pretrained CNN model(ResNet50, VGG, etc) and use the features with a CTC loss function.
I want to build it as a text recognition model.
Anyone on how can i achieve this ?
Adesh Gautam
- 51
- 7
0
votes
1 answer
Can prefix beam search commonly used in speech recognition with CTC be implemented in such a simpler way?
I am learning about speech recognition recently, and I have learned that the idea of prefix beam search is to merge paths with the same prefix, such as [1,1,_] and [_,1,_] (as you can see, _ indicates blank mark).
Based on this understanding, I…
Randool
- 11
- 2
0
votes
1 answer
CTC + BLSTM Architecture Stalls/Hangs before 1st epoch
I am working on a code which recognizes online handwriting recognition.
It works with CTC loss function and Word Beam Search (custom implementation: githubharald)
TF Version: 1.14.0
Following are the parameters used:
batch_size: 128
total_epoches:…
mastershot201
- 3
- 1
0
votes
1 answer
Negative labels withh keras ocr example
I am trying to implement an handwriting ocr based on the keras ocr example: link.
However I get the following error:
InvalidArgumentError: All labels must be nonnegative integers, batch: 0 labels:…
chalulu
- 21
- 1
0
votes
1 answer
How to convert SparseToDense's to Tensor or int array?
I have implemented LSTM-CTC model in android device and comes to this code block.
SparseToDense sparseDense = ctc_decode(out, input_length, true, beam_width, top_paths);
I am getting the sparseDense object perfectly with expected shape but i don't…
Mayur Kanojiya
- 158
- 1
- 8
0
votes
1 answer
CTC model does not learn
I am trying to program a Keras model for audio transcription using connectionist temporal classification. Using a mostly working framewise classification model and the OCR example, I came up with the model given below, which I want to train on…
Anaphory
- 6,045
- 4
- 37
- 68