Question about pretrained "adda_deeplab_v3p.h5"

[GWwangshuo]

Could you please offer the pretrained model adda_deeplab_v3p.h5 ? Thanks a lot.

Moreover, it seems that the commands of training G and training G and D are the same. Are these training commands correct or not?

• training G
  python.exe ./train_adda_seg.py ./data/inria_test source_image source_label_index target_image adda_deeplab_v3p.h5 --optimizer adam --base_learning_rate 1e-4 --min_learning_rate 1e-7 --image_width 256 --image_height 256 --image_channel 3 --image_suffix .png --label_suffix .png --n_class 2 --batch_size 2 --iterations 50 --weight_decay 1e-4 --initializer he_normal --bn_epsilon 1e-3 --bn_momentum 0.99 --pre_trained_model ./logs/checkpoints/deeplab_v3p_base.h5 --source_fname_file ./data/inria_test/source.txt --target_fname_file ./data/inria_test/target.txt --logs_dir ./logs --augmentations flip_x,flip_y,random_crop --display 1 --snapshot 5
• training G and D
  python.exe ./train_adda_seg.py ./data/inria_test source_image source_label_index target_image adda_deeplab_v3p.h5 --optimizer adam --base_learning_rate 1e-4 --min_learning_rate 1e-7 --image_width 256 --image_height 256 --image_channel 3 --image_suffix .png --label_suffix .png --n_class 2 --batch_size 2 --iterations 50 --weight_decay 1e-4 --initializer he_normal --bn_epsilon 1e-3 --bn_momentum 0.99 --pre_trained_model ./logs/checkpoints/deeplab_v3p_base.h5 --source_fname_file ./data/inria_test/source.txt --target_fname_file ./data/inria_test/target.txt --logs_dir ./logs --augmentations flip_x,flip_y,random_crop --display 1 --snapshot 5