Frédéric Branchaud-Charron
Frédéric Branchaud-Charron
Yeah I see the same thing. Using the functional API would make it simpler as well. Extracting a class Tiramisu would make the code cleaner as well. Since only the...
Could it be the batch_size? They use 3 in the paper. Also, they are doing a fully Convolutional Network. You could get more info [here](https://github.com/SimJeg/FC-DenseNet/blob/master/FC-DenseNet.py)
Have you looked at postprocess in net/yolo/test? There is a _tresh dict that may disrupt your output. I had to remove it to make it work
I think you should inherit ImageDataGenerator and apply your modifications in the child class.
Are you on the latest version of Keras? I think we fixed the behaviour a while ago.
Hello, Unfortunately, I'm not affiliated with Keras anymore. You might want to reach out to @fchollet. This project is now deprecated and we should use `tf.data` or Keras Preprocessing layers.
I think this has been fixed on Keras master? ________________________________ De : shuangyangwang Envoyé : 6 novembre 2018 21:42:44 À : keras-team/keras-preprocessing Cc : Subscribed Objet : Re: [keras-team/keras-preprocessing] TimeseriesGenerator...
There is no particular reason, would using the `logging` module good enough? You could disable the logs with an ENV var.
Yes we can do that. PR welcome.
To avoid breaking changes, we should document this behaviour.