search

HasnainRaz / FC-DenseNet-TensorFlow

Fully Convolutional DenseNet (A.K.A 100 layer tiramisu) for semantic segmentation of images implemented in TensorFlow.
MIT License
123 stars 41 forks source link

normalize_data problem why image/255 and mask/255 when training #17

Closed sdjsngs closed 5 years ago

sdjsngs commented 5 years ago

before training, why should image/255 and mask/255 my mask's each pixel is an index in [0,classnumber] if mask/255 they will all be 0 (class:background) when i use pascal VOC 2012 dataset
one more thing why use 30 in parse resize normaliza and shuffle thanks a lot

HasnainRaz commented 5 years ago

You can comment out the normalization function in the dataloader. It was written for binary class segmentation problems where the mask has [0, 255] values.

The 30 is just a number I put in, you can increase the buffer size as you wish. There is not restriction on it.