DL之CycleGAN：基于TF利用CycleGAN模型对apple2orange数据集实现图像转换—训练&测试过程全记录

DL之CycleGAN：基于TF利用CycleGAN模型对apple2orange数据集实现图像转换—训练&测试过程全记录

---

apple2orange数据集

1、数据集下载：https://people.eecs.berkeley.edu/~taesung_park/CycleGAN/datasets/apple2orange.zip

2、数据集展示

输出结果

1、SCALARS

输入与输出
Learning rate、Loss

2、IMAGES

3、GRAPHS

4、DISTRIBUTIONS

训练&测试过程全记录

1、

2018-10-09 22:02:51.396724: W tensorflow/core/framework/allocator.cc:113] Allocation of 822083584 exceeds 10% of system memory.
2018-10-09 22:02:51.396724: W tensorflow/core/framework/allocator.cc:113] Allocation of 822083584 exceeds 10% of system memory.
2018-10-09 22:03:00.267249: W tensorflow/core/framework/allocator.cc:113] Allocation of 822083584 exceeds 10% of system memory.
2018-10-09 22:03:03.406296: W tensorflow/core/framework/allocator.cc:113] Allocation of 822083584 exceeds 10% of system memory.
2018-10-09 22:03:04.851590: W tensorflow/core/framework/allocator.cc:113] Allocation of 822083584 exceeds 10% of system memory.

第一次
INFO:root:-----------Step 0:-------------
INFO:root:  G_loss   : 16.109222412109375
INFO:root:  D_Y_loss : 2.2684788703918457
INFO:root:  F_loss   : 16.630470275878906
INFO:root:  D_X_loss : 2.722561836242676
INFO:root:Model saved in file: checkpoints/20181009-2201/model.ckpt-0

……

第二次  2018-10-09-24：00 ~ 10-10-07-30
INFO:root:-----------Step 0:-------------
INFO:root:  G_loss   : 15.09631061553955
INFO:root:  D_Y_loss : 2.054762601852417
INFO:root:  F_loss   : 15.429546356201172
INFO:root:  D_X_loss : 1.5412709712982178
INFO:root:Model saved in file: checkpoints/20181009-2349/model.ckpt-0
INFO:root:-----------Step 100:-------------
INFO:root:  G_loss   : 11.179760932922363
INFO:root:  D_Y_loss : 0.2794855833053589
INFO:root:  F_loss   : 11.133889198303223
INFO:root:  D_X_loss : 0.4493861794471741
INFO:root:-----------Step 200:-------------
INFO:root:  G_loss   : 7.770391464233398
INFO:root:  D_Y_loss : 0.21000759303569794
INFO:root:  F_loss   : 7.941715717315674
INFO:root:  D_X_loss : 0.20347760617733002
INFO:root:-----------Step 300:-------------
INFO:root:  G_loss   : 6.762332439422607
INFO:root:  D_Y_loss : 0.16613411903381348
INFO:root:  F_loss   : 6.651090621948242
INFO:root:  D_X_loss : 0.17436569929122925
……