search

dBeker / Faster-RCNN-TensorFlow-Python3

Tensorflow Faster R-CNN for Windows/Linux and Python 3 (3.5/3.6/3.7)
MIT License
612 stars 329 forks source link

Running train.py gives no output #88

Closed grignarder closed 5 years ago

grignarder commented 5 years ago

I can't find any output weights file after i run train.py. Should I add some parameters when python train.py or just use python train.py?

Since no output is given, I met problems when trying to run demo.py. Thanks a lot!

morpheusthewhite commented 5 years ago

No extra parameter is needed when running python train.py.

What problems do you meet when running demo.py?

grignarder commented 5 years ago

(tensorflow) C:\Faster-RCNN-TensorFlow-Python3>python demo.py output\res101\voc_2007_trainval+voc_2012_trainval\default\res101_faster_rcnn_iter_110000.ckpt Traceback (most recent call last): File "demo.py", line 123, in 'our server and place them properly?').format(tfmodel + '.meta')) OSError: output\res101\voc_2007_trainval+voc_2012_trainval\default\res101_faster_rcnn_iter_110000.ckpt.meta not found. Did you download the proper networks from our server and place them properly?

The output is like this. Is there anything that should be changed in the demo.py? Sorry to trouble you.

grignarder commented 5 years ago

By the way, i modified the maximum iteration time to 100. I don't think it contributes to this problem.

morpheusthewhite commented 5 years ago

Yes, the checkpoint is saved every 5000 iterations, so if you reduce the iterations to 100 no checkpoint will be saved.

You can change the number of iterations necessary to save the checkpoint in lib/config/config.py (the variable is defined at line 44)