There is no Pascal's pre-trained model

[makalo]

Hi,I know about you load the pre-trained model of pascal-VOC in your thesis.But you just provide the 'CHIP_PGN' model .Could you provide a tensorflow version of Pascal's pre-training model? I don't know if your Pascal model is transferred from caffe. If so, can you provide a conversion method?

[Engineering-Course]

We used the converted pre-trained model, deeplab_resnet_init.ckpt, provided by Deeplab v2

[makalo]

I start training from the pre-trained model mentioned above on the Pascal-Person-Part dataset, but only get mIoU around 66.5, which is much lower than 68.4 reported in the paper. I wonder why the gap exists, and could you please give me your well-trained model on this dataset.

[tengteng95]

@Engineering-Course Hi, I'm also conduting experiments on PASCAL-PERSON-PART dataset, could you please share your pre-trained weights for this dataset! Thanks a lot.

[Engineering-Course]

You can train the model on PASCAL dataset, cause we have released the code and implementation details. We have the multi-scale and fliping inference to reach the results in the paper.

[QHan1st]

@Engineering-Course Hello,I'm training the model onPASCAL-PERSON-PART dataset,could you tell me how many changes I should do in the script concretly?Thank you.

[QHan1st]

@makalo Hello,I also train the model on the Pascal-Person-Part dataset,Could you tell mehow many changes you did in the script?Thank you.