local variable 'model_files' referenced before assignment

[alic-xc]

Hi guys,
Having successfully training my model for the past 1 week. i have a problem trying to evaluate them.

import pixellib
from pixellib.custom_train import instance_custom_training


train_maskrcnn = instance_custom_training()
train_maskrcnn.modelConfig(network_backbone = "resnet50", num_classes= 1)
train_maskrcnn.load_dataset("fingerlings")
train_maskrcnn.evaluate_model("mask_rccn_models")

This is the error

Using resnet50 as network backbone For Mask R-CNN model
---------------------------------------------------------------------------
UnboundLocalError                         Traceback (most recent call last)
<ipython-input-20-55777605efed> in <module>()
      6 train_maskrcnn.modelConfig(network_backbone = "resnet50", num_classes= 1)
      7 train_maskrcnn.load_dataset("fingerlings")
----> 8 train_maskrcnn.evaluate_model("mask_rccn_models")

/usr/local/lib/python3.7/dist-packages/pixellib/custom_train.py in evaluate_model(self, model_path, iou_threshold)
    132         if os.path.isdir(model_path):
    133             model_files = sorted([os.path.join(model_path, file_name) for file_name in os.listdir(model_path)])
--> 134 
    135         for modelfile in model_files:
    136             if str(modelfile).endswith(".h5"):

UnboundLocalError: local variable 'model_files' referenced before assignment

I need your help. will appreciate any effort

[ayoolaolafenwa]

@alic-xc you used the wrong name to reference the path to the saved models.

train_maskrcnn.evaluate_model("mask_rccn_models")

The name of the default path to saved models is mask_rcnn_models not mask_rccn_models.

Confirm by checking the name of the path to saved models in your directory, you will realize that the path name does not match the name referenced in your code.

Correct code corresponding to the default path of the saved models:

train_maskrcnn.evaluate_model("mask_rcnn_models")

Note: You can change the name of the models' path to any name you want, but ensure that you reference that exact name given in your code.

[alic-xc]

@ayoolaolafenwa Sure. Thanks. Working now.

[MarkWijkhuizen]

I encountered the same problem. The reason for the typo is the typo in the awesome tutorial Custom Instance Segmentation Training With 7 Lines Of Code.

Would prevent a lot of frustrations if this typo could be corrected 😃