image-segmentation-keras

image-segmentation-keras copied to clipboard

How can we resume training from the last checkpoint saved? 2nd Question, I've images of 4 different channels (Red, green, NIR, Red edge) in 4 different folders, each folder contain...

mrafifaisal

Hi, Thanks for the good work. Can I ask for a python script to convert coco2017 dataset into the format supported by this repository. I want to train model using...

patelajaychh

loss: nan - acc: nan - val_loss: nan - val_acc: nan Why

1

python train.py --save_weights_path=weights/ex1 --train_images="data/dataset1/images_prepped_train/" --train_annotations="data/dataset1/annotations_prepped_train/" --val_images="data/dataset1/images_prepped_test/" --val_annotations="data/dataset1/annotations_prepped_test/" --n_classes=10 --input_height=320 --input_width=640 --model_name="vgg_segnet" i follow what the tutorial,but the result is this: Model output shape (None, 51200, 10) Epoch 1/1 Epoch 1/1...

qq191513

ImportError: attempted relative import with no known parent package

1

I'm attempting to train a unet model, and am getting this error: Traceback (most recent call last): File "/home/kseg/keras_segmentation/train.py", line 4, in from .data_utils.data_loader import image_segmentation_generator, \ ImportError: attempted relative...

bc-bytes

poor results with gray scale images

1

Hi, we trained unet_vgg19 model using GT with training accuracy of 99.6 % on grayscale (converted RGB) data . After getting such a good accuracies in training getting .60 as...

jaigsingla

Grayscale PNG images wrong assertion on prediction

1

When loading grayscale images they have shape=2, whick causes bellow assertion to fire in _predict.py::predict()_: assert (len(inp.shape) == 3 or len(inp.shape) == 1 or len(inp.shape) == 4), "Image should be...

DmitryMalishev

I have made binary classification using vgg_unet, now I want to plot a ROC-curve for accuracy evaluation. How can I get the map that indicates the probability value of each...

yyruilin

Pixel value issue

1

 I am keeping getting this error. Do you know why this happened ?

WEIYI2021

How do I use PNG pictures with α channel to train? I have changed the parameters in train.py (read_image_type=-1), but it does not work. OP_REQUIRES failed at conv_ops.cc:529 : Invalid...

LaBiXiaoChai

Ignore background class

3

I have a **background class** masked as **'0'**. I do not want to train it so I use flag '**ignore_zero_class**' but the accuracy doesn't increase beyond 3%.

silvernox-dev