NelsonGon comments

Results 21 comments of


                                            NelsonGon

my predicted image is gray image

I have partially solved this(for now) although I still get wrong predictions sometimes. You can try to see if [it](https://github.com/Nelson-Gon/unet/) works for you or if you can improve it.

google.protobuf.text_format.ParseError: 38:80 : 'layers { bottom: 'd3c' bottom: 'u3a' top: 'd3cc' name: 'crop_d3c-d3cc' type: CROP }': Enum type "caffe.V1LayerParameter.LayerType" has no value named CROP.

I think this is the wrong repository. As far as I am aware, this does not use any caffemodels. Perhaps you meant to open an issue at the original authors'...

The program reported an error when I ran it on colab

I have compiled a list of frequently asked questions that I have encountered. You can find it [here](https://github.com/Nelson-Gon/cytounet/wiki). Hope it helps!

Error when checking target: expected conv2d_134 to have 4 dimensions, but got array with shape (24, 320, 320)

Could you add a sample of your code? This is probably to do with using the wrong input shape or using a pretrained model whose weights expect different shapes. Hard...

I'm trying to detect this crack but can't, can't detect it at all, what can I do to improve the prime

Do you have some sample images with their ground truths?

ValueError: Error when checking target: expected conv2d_11 to have shape (16, 16, 1) but got array with shape (256, 256, 1)

You need to reshape your images to the target size (256,256) or change the target size. You can try to use my [fork](https://github.com/Nelson-Gon/unet) and see if it works for you.

TypeError: ufunc 'true_divide' output (typecode 'd') could not be coerced to provided output parameter (typecode 'B') according to the casting rule ''same_kind''

You should avoid using `/= 255`. I have solved this [here](https://github.com/Nelson-Gon/unet). Note that this is a work in progress so it may fail. However, your error is to do with...

NelsonGon