Training Accuracy is always 1.00 and the Minibatch error is always 0.0%
ChristmasLatte opened this issue · 3 comments
Hi, I have a trouble when i was training model.The minibatch loss seems normal, but the training accuracy is always 1 and minibatch error is always 0.0%.
And I just want extract buildings from image, and my mask label channels is 3,should i set n_class=3 ?
Here is my code:
from tf_unet import unet, util, image_util
data_provider = image_util.ImageDataProvider("data/train/*.tif")
net = unet.Unet(layers=3, features_root=64, channels=3, n_class=3)
trainer = unet.Trainer(net)
path = trainer.train(data_provider, "./data/unet_trained_bgs_example_data", training_iters=32, epochs=100, dropout=0.5)
verification
...
data_provider = image_util.ImageDataProvider("data/test/*.tif")
x_test, y_test = data_provider(1)
prediction = net.predict("./data/unet_trained_bgs_example_data/model.ckpt", x_test)
unet.error_rate(prediction, util.crop_to_shape(y_test, prediction.shape))
img = util.combine_img_prediction(x_test, y_test, prediction)
util.save_image(img, "prediction.jpg")
Hi @ChristmasLatte , hard to tell whats going on from what I see here.
I'm currently working on a re-implementation of tf_unet
that is based on tensorflow 2 & keras and addressing some of the issues tf_unet had.
Maybe this could help you:
https://github.com/jakeret/unet
Okay, thanks for your answer @jakeret .I will refre another projected.
Sorry i am a beginner in u-net, I also have a little question that if i want to class an image into foreground and background, should i set n_class 1?
I would appreciate it if you have time to reply.
Hi @ChristmasLatte, I am experiencing the same problem and was wondering if you managed to solve it? Thank you.