About calculating "accuracy_class" in the "train/train_val.prototxt"

ShapeNet / RenderForCNN

Render for CNN: Viewpoint Estimation in Images Using CNNs Trained with Rendered 3D Model Views

Other

248 stars 82 forks source link

Closed rockstone533 closed 8 years ago

rockstone533 commented 8 years ago

Why shall we compute "accuracy_class" by the bottom layer be the same "label_class"?

refer to the code in the train/train_val.prototxt:

layers {
  name: "accuracy_class"
  type: ACCURACY
  bottom: "label_class"
  bottom: "label_class"
  top: "accuracy_class"
}

charlesq34 commented 8 years ago

we can use SilenceLayer as well - it just prevents label_class to be printed in the log file, since it hasn't been used anywhere else.