Closed delray closed 7 years ago
Shoot, sorry about this man. I think I know what the problem is. Did you happen to train your model, call git pull, and then run the test script? If that's the case, I think that some changes I've made in model.py and utils.py may have broken this.
From here you have two options:
model.py
and utils.py
files with the ones that are listed here: https://github.com/jmiller656/EDSR-Tensorflow/tree/4280fefa95d1f0077d8682f6a0e74cb8724139cb
and then try testing again. Hope this helps!
Thats exactly what I did - retraining now, ill let you know how it goes...thanks for all this help...
No prob!
Thanks for the test.py script, but getting some errors here, it looks like the weights aren't being found in the checkpoint during the restore process. I'm using python 3.5.2 and tensorflow 1.1.0 (gpu). I'm happy to try out some other things to try and get this working. I tried inserting a 'tf.reset_default_graph()' before the restore process, but to no avail. When training, the only thing i changed from the defaults were --dataset and --imgsize here is my train cmd used:
python3 test.py --dataset /media/sugbuntu/Data/TSA/data/output/stage1_aps_front_crop/ --imgsize 128