CheXNet-Keras

CheXNet-Keras copied to clipboard

Predictions very low (p<< 0.5)

5

Hello, After running python test.py I found reasonable result (~0.80) but all the predictions are very low (

gmarceca

can't load pretrained model CheXNet_Keras_0.3.0_weights.h5 into Keras Densenet121

1

Here I have asked the question https://stackoverflow.com/questions/64390544/unable-to-load-chexnet-pre-trained-weight-file-to-densenet121 `` densenet = tf.keras.applications.DenseNet121( include_top=False, weights="CheXNet_Keras_0.3.0_weights.h5", input_shape=(224,224,3) ) `` ValueError: You are trying to load a weight file containing 242 layers into a...

WrathofBhuvan11

Problem with ModelFactory class while loading ChexNet-keras.h5 pretrained file

1

ValueError: You are trying to load a weight file containing 242 layers into a model with 241 layers.

WrathofBhuvan11

Can some please tell order of the class name for the predicted class index , from pre trained weights -https://drive.google.com/file/d/19BllaOvs2x5PLV_vlWMy4i8LapLb2j6b/view or how to know it. I could get a prediction...

DineshKA87

Thought this might help people get started quickly.

Ouwen

MichelleVivita

Have you tried with all models (VGG16, VGG19, DenseNet121, ResNet50, InceptionV3, InceptionResNetV2, NASNetMobile, NASNetLarge) and can you give me the results of those?

hmchuong

MLikelihood

The implementation of learning rate decay every time validation loss stop decrease is vary elegant in the code, but unfortunately I use native tensorflow to implemente CheXNet instead of Keras,...

chasewinds

in some cases validation loss can be high yet with better mean auc.current code copies weights.h5 (which are monitored according to val_loss.) to best_weights which is a bug.best weights should...

magomagic