Get "None" when try to evaluate

Hi, running an issue when trying to evaluate the detections in None any suggestion?

(keras) C:\Users\z\Anaconda3\envs\keras\Lib\site-packages\keras_rnet\keras_retinanet>python bin/evaluate.py csv valc.csv  class.csv D:\Dropbox\DL\open_image\data\resnet50_csv_01.h5
Using TensorFlow backend.
2019-01-06 03:30:17.319312: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2
Loading model, this may take a second...
Running network: 100% (1 of 1) |#################################################| Elapsed Time: 0:00:11 Time:  0:00:11
Parsing annotations: 100% (1 of 1) |#############################################| Elapsed Time: 0:00:00 Time:  0:00:00
label 0
detections None
Traceback (most recent call last):
  File "bin/evaluate.py", line 188, in <module>
    main()
  File "bin/evaluate.py", line 167, in main
    save_path=args.save_path
  File "bin\..\..\keras_retinanet\utils\eval.py", line 197, in evaluate
    for d in detections:
TypeError: 'NoneType' object is not iterable

(keras) C:\Users\z\Anaconda3\envs\keras\Lib\site-packages\keras_rnet\keras_retinanet>

[dabasmoti]

Thanks. Great repository!

[hgaiser]

If you start from 1, the network will “train” with an additional class (index 0), for which there is no data available. The evaluator should handle this more gracefully, the network is likely trained properly.