search

aurora95 / Keras-FCN

Keras-tensorflow implementation of Fully Convolutional Networks for Semantic Segmentation（Unfinished）
MIT License
650 stars 268 forks source link

Mean IOU #32

Open SuianHan opened 7 years ago

SuianHan commented 7 years ago

I've run train.py with all the default setting on PASCAL VOC dataset with AtrousFCN_Resnet50_16s with a epoch number of 250. the result is not satisfactory with a Mean IOU of 0.24 and the segmented pictures generated by the evaluate.py are barely recognized. I've run both train.py and evaluate.py on a nvidia TITAN X with 12G memory. What is the problem? I've also run train.py with other nets like VGG16 or Resnet50_32s. The results are even worse.

XChuanLee commented 6 years ago

excuse me.. I also met this problem.. Have you solve it??

SuianHan commented 6 years ago

@IronDroid2088 I didn't try it on PASCAL VOC anymore. However, by using the model(Resnet50) on other data-sets, I get decent result with bigger training batch size and epoch number. Therefore, i suppose you may achieve better result on PASCAL with increasing these two parameters.