Matthew Avaylon

Metric changes

2

Is there a way to change the metric from accuracy to a custom metric like iou? I changed train.py to have model.compile(loss=loss_k, optimizer=optimizer_name, metrics=[iou]) #replaced accuracy with iou where i...

Why do we have the reshaping to (width*height, channels) in the first place? I understand the it is reshaped to that both before and after the model. But why?

1

Why do we have the reshaping to (width*height, channels) in the first place? I understand the it is reshaped to that both before and after the model. But why? _Originally...

What's the reasoning behind reshaping the masks and then the model output to (width*height, channel)? When you run predictions you change back: pr = pr.reshape((output_height, output_width, n_classes)).argmax(axis=2). Everywhere else I...

ResNet50

1

Hi I've been using your Resnet50 for a while now. I looked at it and saw that there was one call of the "one_side_pad" function. I know this is used...

Why is it the dimension of the mask as follows in data_utils.py seg_labels = np.reshape(seg_labels, (width*height, nClasses)) Why isnt it (batch size, widht, height, nclasses)

I was looking at "image_segmentation_generator" in the data_loader.py and printed out the shape of the individual output image and mask shapes. I got: img shape (416, 608, 3) seg_shape (252928,...

Will this be update to use .fit instead of fit_generator? I tried and only works when I keep it at fit_generator.

## Motivation What was the reasoning behind this change? Please explain the changes briefly. Adding ExternalResource from hdmf to pynwb. ## How to test the behavior? ``` Unit Tests ##...

Keras from Tensorflow Version

1

Will there be an update where it uses keras from tensorflow and not keras the api?

Using the CRF_Layer results is Seg Fault 11

1

I'm trying to create a UNet with the CRF Layer as the output layer but when I run it through terminal (Mac OS Mojave 10.14.6), the images are found and...