Error(s) in loading state_dict for PatchcoreLightning

[Luksalos]

Describe the bug

Running tools/inference/torch_inference.py throws:

Traceback (most recent call last):
  File "/workspace/tools/inference/torch_inference.py", line 97, in <module>
    infer()
  File "/workspace/tools/inference/torch_inference.py", line 73, in infer
    inferencer = TorchInferencer(config=args.config, model_source=args.weights)
  File "/workspace/anomalib/deploy/inferencers/torch_inferencer.py", line 54, in __init__
    self.model = self.load_model(model_source)
  File "/workspace/anomalib/deploy/inferencers/torch_inferencer.py", line 85, in load_model
    model.load_state_dict(torch.load(path)["state_dict"])
  File "/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1497, in load_state_dict
    raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for PatchcoreLightning:
	Unexpected key(s) in state_dict: "normalization_metrics.min", "normalization_metrics.max".

To Reproduce

1. Using default patchcore/config.yaml
2. Train the model: python tools/train.py --config anomalib/models/patchcore/config.yaml
3. Use trained model for torch_inference:

python tools/inference/torch_inference.py \
--config anomalib/models/patchcore/config.yaml \
--weights results/patchcore/mvtec/bottle/weights/model-v14.ckpt \
--input datasets/MVTec/bottle/test/broken_large/000.png \
--output results/patchcore/mvtec/bottle/images

Additional context

Error is present since commit d6951eb

[djdameln]

@ashwinvaidya17 can you have a look at this? It seems to be related to the normalization redesign

[alexriedel1]

The normalization callback setup is not called before the model state dict loading in torch_inference.py

[alevangel]

I'm having a simlar issue, here the runtime error:

RuntimeError: Error(s) in loading state_dict for PatchcoreLightning:
	Missing key(s) in state_dict: "model.feature_extractor.feature_extractor.conv1.weight", "model.feature_extractor.feature_extractor.bn1.weight", "model.feature_extractor.feature_extractor.bn1.bias", "model.feature_extractor.feature_extractor.bn1.running_mean", "model.feature_extractor.feature_extractor.bn1.running_var", "model.feature_extractor.feature_extractor.layer1.0.conv1.weight", "model.feature_extractor.feature_extractor.layer1.0.bn1.weight", "model.feature_extractor.feature_extractor.layer1.0.bn1.bias", "model.feature_extractor.feature_extractor.layer1.0.bn1.running_mean", "model.feature_extractor.feature_extractor.layer1.0.bn1.running_var", "model.feature_extractor.feature_extractor.layer1.0.conv2.weight", "model.feature_extractor.feature_extractor.layer1.0.bn2.weight", "model.feature_extractor.feature_extractor.layer1.0.bn2.bias", "model.feature_extractor.feature_extractor.layer1.0.bn2.running_mean", "model.feature_extractor.feature_extractor.layer1.0.bn2.running_var", "model.feature_extractor.feature_extractor.layer1.0.conv3.weight", "model.feature_extractor.feature_extractor.layer1.0.bn3.weight", "model.feature_extractor.feature_extractor.layer1.0.bn3.bias", "model.feature_extractor.feature_extractor.layer1.0.bn3.running_mean", "model.feature_extractor.feature_extractor.layer1.0.bn3.running_var", "model.feature_extractor.feature_extractor.layer1.0.downsample.0.weight", "model.feature_extractor.feature_extractor.layer1.0.downsample.1.weight", "model.feature_extractor.feature_extractor.layer1.0.downsample.1.bias", "model.feature_extractor.feature_extractor.layer1.0.downsample.1.running_mean", "model.feature_extractor.feature_extractor.layer1.0.downsample.1.running_var", "model.feature_extractor.feature_extractor.layer1.1.conv1.weight", "model.feature_extractor.feature_extractor.layer1.1.bn1.weight", "model.feature_extractor.feature_extractor.layer1.1.bn1.bias", "model.feature_extractor.feature_extractor.layer1.1.bn1.running_mean", "model.feature_extractor.feature_extractor.layer1.1.bn1.running_var", "model.feature_extractor.feature_extractor.layer1.1.conv2.weight", "model.feature_extractor.feature_extractor.layer1.1.bn2.weight", "model.feature_extractor.feature_extractor.layer1.1.bn2.bias", "model.feature_extractor.feature_extractor.layer1.1.bn2.running_mean", "model.feature_extractor.feature_extractor.layer1.1.bn2.running_var", "model.feature_extractor.feature_extractor.layer1.1.conv3.weight", "model.feature_extractor.feature_extractor.layer1.1.bn3.weight", "model.feature_extractor.feature_extractor.layer1.1.bn3.bias", "model.feature_extractor.feature_extractor.layer1.1.bn3.running_mean", "model.feature_extractor.feature_extractor.layer1.1.bn3.running_var", "model.feature_extractor.feature_extractor.layer1.2.conv1.weight", "model.feature_extractor.feature_extractor.layer1.2.bn1.weight", "model.feature_extractor.feature_extractor.layer1.2.bn1.bias", "model.feature_extractor.feature_extractor.layer1.2.bn1.running_mean", "model.feature_extractor.feature_extractor.layer1.2.bn1.running_var", "model.feature_extractor.feature_extractor.layer1.2.conv2.weight", "model.feature_extractor.feature_extractor.layer1.2.bn2.weight", "model.feature_extractor.feature_extractor.layer1.2.bn2.bias", "model.feature_extractor.feature_extractor.layer1.2.bn2.running_mean", "model.feature_extractor.feature_extractor.layer1.2.bn2.running_var", "model.feature_extractor.feature_extractor.layer1.2.conv3.weight", "model.feature_extractor.feature_extractor.layer1.2.bn3.weight", "model.feature_extractor.feature_extractor.layer1.2.bn3.bias", "model.feature_extractor.feature_extractor.layer1.2.bn3.running_mean", "model.feature_extractor.feature_extractor.layer1.2.bn3.running_var", "model.feature_extractor.feature_extractor.layer2.0.conv1.weight", "model.feature_extractor.feature_extractor.layer2.0.bn1.weight", "model.feature_extractor.feature_extractor.layer2.0.bn1.bias", "model.feature_extractor.feature_extractor.layer2.0.bn1.running_mean", "model.feature_extractor.feature_extractor.layer2.0.bn1.running_var", "model.feature_extractor.feature_extractor.layer2.0.conv2.weight", "model.feature_extractor.feature_extractor.layer2.0.bn2.weight", "model.feature_extractor.feature_extractor.layer2.0.bn2.bias", "model.feature_extractor.feature_extractor.layer2.0.bn2.running_mean", "model.feature_extractor.feature_extractor.layer2.0.bn2.running_var", "model.feature_extractor.feature_extractor.layer2.0.conv3.weight", "model.feature_extractor.feature_extractor.layer2.0.bn3.weight", "model.feature_extractor.feature_extractor.layer2.0.bn3.bias", "model.feature_extractor.feature_extractor.layer2.0.bn3.running_mean", "model.feature_extractor.feature_extractor.layer2.0.bn3.running_var", "model.feature_extractor.feature_extractor.layer2.0.downsample.0.weight", "model.feature_extractor.feature_extractor.layer2.0.downsample.1.weight", "model.feature_extractor.feature_extractor.layer2.0.downsample.1.bias", "model.feature_extractor.feature_extractor.layer2.0.downsample.1.running_mean", "model.feature_extractor.feature_extractor.layer2.0.downsample.1.running_var", "model.feature_extractor.feature_extractor.layer2.1.conv1.weight", "model.feature_extractor.feature_extractor.layer2.1.bn1.weight", "model.feature_extractor.feature_extractor.layer2.1.bn1.bias", "model.feature_extractor.feature_extractor.layer2.1.bn1.running_mean", "model.feature_extractor.feature_extractor.layer2.1.bn1.running_var", "model.feature_extractor.feature_extractor.layer2.1.conv2.weight", "model.feature_extractor.feature_extractor.layer2.1.bn2.weight", "model.feature_extractor.feature_extractor.layer2.1.bn2.bias", "model.feature_extractor.feature_extractor.layer2.1.bn2.running_mean", "model.feature_extractor.feature_extractor.layer2.1.bn2.running_var", "model.feature_extractor.feature_extractor.layer2.1.conv3.weight", "model.feature_extractor.feature_extractor.layer2.1.bn3.weight", "model.feature_extractor.feature_extractor.layer2.1.bn3.bias", "model.feature_extractor.feature_extractor.layer2.1.bn3.running_mean", "model.feature_extractor.feature_extractor.layer2.1.bn3.running_var", "model.feature_extractor.feature_extractor.layer2.2.conv1.weight", "model.feature_extractor.feature_extractor.layer2.2.bn1.weight", "model.feature_extractor.feature_extractor.layer2.2.bn1.bias", "model.feature_extractor.feature_extractor.layer2.2.bn1.running_mean", "model.feature_extractor.feature_extractor.layer2.2.bn1.running_var", "model.feature_extractor.feature_extractor.layer2.2.conv2.weight", "model.feature_extractor.feature_extractor.layer2.2.bn2.weight", "model.feature_extractor.feature_extractor.layer2.2.bn2.bias", "model.feature_extractor.feature_extractor.layer2.2.bn2.running_mean", "model.feature_extractor.feature_extractor.layer2.2.bn2.running_var", "model.feature_extractor.feature_extractor.layer2.2.conv3.weight", "model.feature_extractor.feature_extractor.layer2.2.bn3.weight", "model.feature_extractor.feature_extractor.layer2.2.bn3.bias", "model.feature_extractor.feature_extractor.layer2.2.bn3.running_mean", "model.feature_extractor.feature_extractor.layer2.2.bn3.running_var", "model.feature_extractor.feature_extractor.layer2.3.conv1.weight", "model.feature_extractor.feature_extractor.layer2.3.bn1.weight", "model.feature_extractor.feature_extractor.layer2.3.bn1.bias", "model.feature_extractor.feature_extractor.layer2.3.bn1.running_mean", "model.feature_extractor.feature_extractor.layer2.3.bn1.running_var", "model.feature_extractor.feature_extractor.layer2.3.conv2.weight", "model.feature_extractor.feature_extractor.layer2.3.bn2.weight", "model.feature_extractor.feature_extractor.layer2.3.bn2.bias", "model.feature_extractor.feature_extractor.layer2.3.bn2.running_mean", "model.feature_extractor.feature_extractor.layer2.3.bn2.running_var", "model.feature_extractor.feature_extractor.layer2.3.conv3.weight", "model.feature_extractor.feature_extractor.layer2.3.bn3.weight", "model.feature_extractor.feature_extractor.layer2.3.bn3.bias", "model.feature_extractor.feature_extractor.layer2.3.bn3.running_mean", "model.feature_extractor.feature_extractor.layer2.3.bn3.running_var", "model.feature_extractor.feature_extractor.layer3.0.conv1.weight", "model.feature_extractor.feature_extractor.layer3.0.bn1.weight", "model.feature_extractor.feature_extractor.layer3.0.bn1.bias", "model.feature_extractor.feature_extractor.layer3.0.bn1.running_mean", "model.feature_extractor.feature_extractor.layer3.0.bn1.running_var", "model.feature_extractor.feature_extractor.layer3.0.conv2.weight", "model.feature_extractor.feature_extractor.layer3.0.bn2.weight", "model.feature_extractor.feature_extractor.layer3.0.bn2.bias", "model.feature_extractor.feature_extractor.layer3.0.bn2.running_mean", "model.feature_extractor.feature_extractor.layer3.0.bn2.running_var", "model.feature_extractor.feature_extractor.layer3.0.conv3.weight", "model.feature_extractor.feature_extractor.layer3.0.bn3.weight", "model.feature_extractor.feature_extractor.layer3.0.bn3.bias", "model.feature_extractor.feature_extractor.layer3.0.bn3.running_mean", "model.feature_extractor.feature_extractor.layer3.0.bn3.running_var", "model.feature_extractor.feature_extractor.layer3.0.downsample.0.weight", "model.feature_extractor.feature_extractor.layer3.0.downsample.1.weight", "model.feature_extractor.feature_extractor.layer3.0.downsample.1.bias", "model.feature_extractor.feature_extractor.layer3.0.downsample.1.running_mean", "model.feature_extractor.feature_extractor.layer3.0.downsample.1.running_var", "model.feature_extractor.feature_extractor.layer3.1.conv1.weight", "model.feature_extractor.feature_extractor.layer3.1.bn1.weight", "model.feature_extractor.feature_extractor.layer3.1.bn1.bias", "model.feature_extractor.feature_extractor.layer3.1.bn1.running_mean", "model.feature_extractor.feature_extractor.layer3.1.bn1.running_var", "model.feature_extractor.feature_extractor.layer3.1.conv2.weight", "model.feature_extractor.feature_extractor.layer3.1.bn2.weight", "model.feature_extractor.feature_extractor.layer3.1.bn2.bias", "model.feature_extractor.feature_extractor.layer3.1.bn2.running_mean", "model.feature_extractor.feature_extractor.layer3.1.bn2.running_var", "model.feature_extractor.feature_extractor.layer3.1.conv3.weight", "model.feature_extractor.feature_extractor.layer3.1.bn3.weight", "model.feature_extractor.feature_extractor.layer3.1.bn3.bias", "model.feature_extractor.feature_extractor.layer3.1.bn3.running_mean", "model.feature_extractor.feature_extractor.layer3.1.bn3.running_var", "model.feature_extractor.feature_extractor.layer3.2.conv1.weight", "model.feature_extractor.feature_extractor.layer3.2.bn1.weight", "model.feature_extractor.feature_extractor.layer3.2.bn1.bias", "model.feature_extractor.feature_extractor.layer3.2.bn1.running_mean", "model.feature_extractor.feature_extractor.layer3.2.bn1.running_var", "model.feature_extractor.feature_extractor.layer3.2.conv2.weight", "model.feature_extractor.feature_extractor.layer3.2.bn2.weight", "model.feature_extractor.feature_extractor.layer3.2.bn2.bias", "model.feature_extractor.feature_extractor.layer3.2.bn2.running_mean", "model.feature_extractor.feature_extractor.layer3.2.bn2.running_var", "model.feature_extractor.feature_extractor.layer3.2.conv3.weight", "model.feature_extractor.feature_extractor.layer3.2.bn3.weight", "model.feature_extractor.feature_extractor.layer3.2.bn3.bias", "model.feature_extractor.feature_extractor.layer3.2.bn3.running_mean", "model.feature_extractor.feature_extractor.layer3.2.bn3.running_var", "model.feature_extractor.feature_extractor.layer3.3.conv1.weight", "model.feature_extractor.feature_extractor.layer3.3.bn1.weight", "model.feature_extractor.feature_extractor.layer3.3.bn1.bias", "model.feature_extractor.feature_extractor.layer3.3.bn1.running_mean", "model.feature_extractor.feature_extractor.layer3.3.bn1.running_var", "model.feature_extractor.feature_extractor.layer3.3.conv2.weight", "model.feature_extractor.feature_extractor.layer3.3.bn2.weight", "model.feature_extractor.feature_extractor.layer3.3.bn2.bias", "model.feature_extractor.feature_extractor.layer3.3.bn2.running_mean", "model.feature_extractor.feature_extractor.layer3.3.bn2.running_var", "model.feature_extractor.feature_extractor.layer3.3.conv3.weight", "model.feature_extractor.feature_extractor.layer3.3.bn3.weight", "model.feature_extractor.feature_extractor.layer3.3.bn3.bias", "model.feature_extractor.feature_extractor.layer3.3.bn3.running_mean", "model.feature_extractor.feature_extractor.layer3.3.bn3.running_var", "model.feature_extractor.feature_extractor.layer3.4.conv1.weight", "model.feature_extractor.feature_extractor.layer3.4.bn1.weight", "model.feature_extractor.feature_extractor.layer3.4.bn1.bias", "model.feature_extractor.feature_extractor.layer3.4.bn1.running_mean", "model.feature_extractor.feature_extractor.layer3.4.bn1.running_var", "model.feature_extractor.feature_extractor.layer3.4.conv2.weight", "model.feature_extractor.feature_extractor.layer3.4.bn2.weight", "model.feature_extractor.feature_extractor.layer3.4.bn2.bias", "model.feature_extractor.feature_extractor.layer3.4.bn2.running_mean", "model.feature_extractor.feature_extractor.layer3.4.bn2.running_var", "model.feature_extractor.feature_extractor.layer3.4.conv3.weight", "model.feature_extractor.feature_extractor.layer3.4.bn3.weight", "model.feature_extractor.feature_extractor.layer3.4.bn3.bias", "model.feature_extractor.feature_extractor.layer3.4.bn3.running_mean", "model.feature_extractor.feature_extractor.layer3.4.bn3.running_var", "model.feature_extractor.feature_extractor.layer3.5.conv1.weight", "model.feature_extractor.feature_extractor.layer3.5.bn1.weight", "model.feature_extractor.feature_extractor.layer3.5.bn1.bias", "model.feature_extractor.feature_extractor.layer3.5.bn1.running_mean", "model.feature_extractor.feature_extractor.layer3.5.bn1.running_var", "model.feature_extractor.feature_extractor.layer3.5.conv2.weight", "model.feature_extractor.feature_extractor.layer3.5.bn2.weight", "model.feature_extractor.feature_extractor.layer3.5.bn2.bias", "model.feature_extractor.feature_extractor.layer3.5.bn2.running_mean", "model.feature_extractor.feature_extractor.layer3.5.bn2.running_var", "model.feature_extractor.feature_extractor.layer3.5.conv3.weight", "model.feature_extractor.feature_extractor.layer3.5.bn3.weight", "model.feature_extractor.feature_extractor.layer3.5.bn3.bias", "model.feature_extractor.feature_extractor.layer3.5.bn3.running_mean", "model.feature_extractor.feature_extractor.layer3.5.bn3.running_var", "model.anomaly_map_generator.blur.kernel". 
	Unexpected key(s) in state_dict: "training_distribution.image_mean", "training_distribution.image_std", "training_distribution.pixel_mean", "training_distribution.pixel_std", "min_max.min", "min_max.max", "model.feature_extractor.backbone.conv1.weight", "model.feature_extractor.backbone.bn1.weight", "model.feature_extractor.backbone.bn1.bias", "model.feature_extractor.backbone.bn1.running_mean", "model.feature_extractor.backbone.bn1.running_var", "model.feature_extractor.backbone.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer1.0.conv1.weight", "model.feature_extractor.backbone.layer1.0.bn1.weight", "model.feature_extractor.backbone.layer1.0.bn1.bias", "model.feature_extractor.backbone.layer1.0.bn1.running_mean", "model.feature_extractor.backbone.layer1.0.bn1.running_var", "model.feature_extractor.backbone.layer1.0.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer1.0.conv2.weight", "model.feature_extractor.backbone.layer1.0.bn2.weight", "model.feature_extractor.backbone.layer1.0.bn2.bias", "model.feature_extractor.backbone.layer1.0.bn2.running_mean", "model.feature_extractor.backbone.layer1.0.bn2.running_var", "model.feature_extractor.backbone.layer1.0.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer1.0.conv3.weight", "model.feature_extractor.backbone.layer1.0.bn3.weight", "model.feature_extractor.backbone.layer1.0.bn3.bias", "model.feature_extractor.backbone.layer1.0.bn3.running_mean", "model.feature_extractor.backbone.layer1.0.bn3.running_var", "model.feature_extractor.backbone.layer1.0.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer1.0.downsample.0.weight", "model.feature_extractor.backbone.layer1.0.downsample.1.weight", "model.feature_extractor.backbone.layer1.0.downsample.1.bias", "model.feature_extractor.backbone.layer1.0.downsample.1.running_mean", "model.feature_extractor.backbone.layer1.0.downsample.1.running_var", "model.feature_extractor.backbone.layer1.0.downsample.1.num_batches_tracked", "model.feature_extractor.backbone.layer1.1.conv1.weight", "model.feature_extractor.backbone.layer1.1.bn1.weight", "model.feature_extractor.backbone.layer1.1.bn1.bias", "model.feature_extractor.backbone.layer1.1.bn1.running_mean", "model.feature_extractor.backbone.layer1.1.bn1.running_var", "model.feature_extractor.backbone.layer1.1.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer1.1.conv2.weight", "model.feature_extractor.backbone.layer1.1.bn2.weight", "model.feature_extractor.backbone.layer1.1.bn2.bias", "model.feature_extractor.backbone.layer1.1.bn2.running_mean", "model.feature_extractor.backbone.layer1.1.bn2.running_var", "model.feature_extractor.backbone.layer1.1.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer1.1.conv3.weight", "model.feature_extractor.backbone.layer1.1.bn3.weight", "model.feature_extractor.backbone.layer1.1.bn3.bias", "model.feature_extractor.backbone.layer1.1.bn3.running_mean", "model.feature_extractor.backbone.layer1.1.bn3.running_var", "model.feature_extractor.backbone.layer1.1.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer1.2.conv1.weight", "model.feature_extractor.backbone.layer1.2.bn1.weight", "model.feature_extractor.backbone.layer1.2.bn1.bias", "model.feature_extractor.backbone.layer1.2.bn1.running_mean", "model.feature_extractor.backbone.layer1.2.bn1.running_var", "model.feature_extractor.backbone.layer1.2.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer1.2.conv2.weight", "model.feature_extractor.backbone.layer1.2.bn2.weight", "model.feature_extractor.backbone.layer1.2.bn2.bias", "model.feature_extractor.backbone.layer1.2.bn2.running_mean", "model.feature_extractor.backbone.layer1.2.bn2.running_var", "model.feature_extractor.backbone.layer1.2.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer1.2.conv3.weight", "model.feature_extractor.backbone.layer1.2.bn3.weight", "model.feature_extractor.backbone.layer1.2.bn3.bias", "model.feature_extractor.backbone.layer1.2.bn3.running_mean", "model.feature_extractor.backbone.layer1.2.bn3.running_var", "model.feature_extractor.backbone.layer1.2.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer2.0.conv1.weight", "model.feature_extractor.backbone.layer2.0.bn1.weight", "model.feature_extractor.backbone.layer2.0.bn1.bias", "model.feature_extractor.backbone.layer2.0.bn1.running_mean", "model.feature_extractor.backbone.layer2.0.bn1.running_var", "model.feature_extractor.backbone.layer2.0.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer2.0.conv2.weight", "model.feature_extractor.backbone.layer2.0.bn2.weight", "model.feature_extractor.backbone.layer2.0.bn2.bias", "model.feature_extractor.backbone.layer2.0.bn2.running_mean", "model.feature_extractor.backbone.layer2.0.bn2.running_var", "model.feature_extractor.backbone.layer2.0.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer2.0.conv3.weight", "model.feature_extractor.backbone.layer2.0.bn3.weight", "model.feature_extractor.backbone.layer2.0.bn3.bias", "model.feature_extractor.backbone.layer2.0.bn3.running_mean", "model.feature_extractor.backbone.layer2.0.bn3.running_var", "model.feature_extractor.backbone.layer2.0.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer2.0.downsample.0.weight", "model.feature_extractor.backbone.layer2.0.downsample.1.weight", "model.feature_extractor.backbone.layer2.0.downsample.1.bias", "model.feature_extractor.backbone.layer2.0.downsample.1.running_mean", "model.feature_extractor.backbone.layer2.0.downsample.1.running_var", "model.feature_extractor.backbone.layer2.0.downsample.1.num_batches_tracked", "model.feature_extractor.backbone.layer2.1.conv1.weight", "model.feature_extractor.backbone.layer2.1.bn1.weight", "model.feature_extractor.backbone.layer2.1.bn1.bias", "model.feature_extractor.backbone.layer2.1.bn1.running_mean", "model.feature_extractor.backbone.layer2.1.bn1.running_var", "model.feature_extractor.backbone.layer2.1.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer2.1.conv2.weight", "model.feature_extractor.backbone.layer2.1.bn2.weight", "model.feature_extractor.backbone.layer2.1.bn2.bias", "model.feature_extractor.backbone.layer2.1.bn2.running_mean", "model.feature_extractor.backbone.layer2.1.bn2.running_var", "model.feature_extractor.backbone.layer2.1.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer2.1.conv3.weight", "model.feature_extractor.backbone.layer2.1.bn3.weight", "model.feature_extractor.backbone.layer2.1.bn3.bias", "model.feature_extractor.backbone.layer2.1.bn3.running_mean", "model.feature_extractor.backbone.layer2.1.bn3.running_var", "model.feature_extractor.backbone.layer2.1.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer2.2.conv1.weight", "model.feature_extractor.backbone.layer2.2.bn1.weight", "model.feature_extractor.backbone.layer2.2.bn1.bias", "model.feature_extractor.backbone.layer2.2.bn1.running_mean", "model.feature_extractor.backbone.layer2.2.bn1.running_var", "model.feature_extractor.backbone.layer2.2.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer2.2.conv2.weight", "model.feature_extractor.backbone.layer2.2.bn2.weight", "model.feature_extractor.backbone.layer2.2.bn2.bias", "model.feature_extractor.backbone.layer2.2.bn2.running_mean", "model.feature_extractor.backbone.layer2.2.bn2.running_var", "model.feature_extractor.backbone.layer2.2.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer2.2.conv3.weight", "model.feature_extractor.backbone.layer2.2.bn3.weight", "model.feature_extractor.backbone.layer2.2.bn3.bias", "model.feature_extractor.backbone.layer2.2.bn3.running_mean", "model.feature_extractor.backbone.layer2.2.bn3.running_var", "model.feature_extractor.backbone.layer2.2.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer2.3.conv1.weight", "model.feature_extractor.backbone.layer2.3.bn1.weight", "model.feature_extractor.backbone.layer2.3.bn1.bias", "model.feature_extractor.backbone.layer2.3.bn1.running_mean", "model.feature_extractor.backbone.layer2.3.bn1.running_var", "model.feature_extractor.backbone.layer2.3.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer2.3.conv2.weight", "model.feature_extractor.backbone.layer2.3.bn2.weight", "model.feature_extractor.backbone.layer2.3.bn2.bias", "model.feature_extractor.backbone.layer2.3.bn2.running_mean", "model.feature_extractor.backbone.layer2.3.bn2.running_var", "model.feature_extractor.backbone.layer2.3.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer2.3.conv3.weight", "model.feature_extractor.backbone.layer2.3.bn3.weight", "model.feature_extractor.backbone.layer2.3.bn3.bias", "model.feature_extractor.backbone.layer2.3.bn3.running_mean", "model.feature_extractor.backbone.layer2.3.bn3.running_var", "model.feature_extractor.backbone.layer2.3.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer3.0.conv1.weight", "model.feature_extractor.backbone.layer3.0.bn1.weight", "model.feature_extractor.backbone.layer3.0.bn1.bias", "model.feature_extractor.backbone.layer3.0.bn1.running_mean", "model.feature_extractor.backbone.layer3.0.bn1.running_var", "model.feature_extractor.backbone.layer3.0.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer3.0.conv2.weight", "model.feature_extractor.backbone.layer3.0.bn2.weight", "model.feature_extractor.backbone.layer3.0.bn2.bias", "model.feature_extractor.backbone.layer3.0.bn2.running_mean", "model.feature_extractor.backbone.layer3.0.bn2.running_var", "model.feature_extractor.backbone.layer3.0.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer3.0.conv3.weight", "model.feature_extractor.backbone.layer3.0.bn3.weight", "model.feature_extractor.backbone.layer3.0.bn3.bias", "model.feature_extractor.backbone.layer3.0.bn3.running_mean", "model.feature_extractor.backbone.layer3.0.bn3.running_var", "model.feature_extractor.backbone.layer3.0.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer3.0.downsample.0.weight", "model.feature_extractor.backbone.layer3.0.downsample.1.weight", "model.feature_extractor.backbone.layer3.0.downsample.1.bias", "model.feature_extractor.backbone.layer3.0.downsample.1.running_mean", "model.feature_extractor.backbone.layer3.0.downsample.1.running_var", "model.feature_extractor.backbone.layer3.0.downsample.1.num_batches_tracked", "model.feature_extractor.backbone.layer3.1.conv1.weight", "model.feature_extractor.backbone.layer3.1.bn1.weight", "model.feature_extractor.backbone.layer3.1.bn1.bias", "model.feature_extractor.backbone.layer3.1.bn1.running_mean", "model.feature_extractor.backbone.layer3.1.bn1.running_var", "model.feature_extractor.backbone.layer3.1.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer3.1.conv2.weight", "model.feature_extractor.backbone.layer3.1.bn2.weight", "model.feature_extractor.backbone.layer3.1.bn2.bias", "model.feature_extractor.backbone.layer3.1.bn2.running_mean", "model.feature_extractor.backbone.layer3.1.bn2.running_var", "model.feature_extractor.backbone.layer3.1.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer3.1.conv3.weight", "model.feature_extractor.backbone.layer3.1.bn3.weight", "model.feature_extractor.backbone.layer3.1.bn3.bias", "model.feature_extractor.backbone.layer3.1.bn3.running_mean", "model.feature_extractor.backbone.layer3.1.bn3.running_var", "model.feature_extractor.backbone.layer3.1.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer3.2.conv1.weight", "model.feature_extractor.backbone.layer3.2.bn1.weight", "model.feature_extractor.backbone.layer3.2.bn1.bias", "model.feature_extractor.backbone.layer3.2.bn1.running_mean", "model.feature_extractor.backbone.layer3.2.bn1.running_var", "model.feature_extractor.backbone.layer3.2.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer3.2.conv2.weight", "model.feature_extractor.backbone.layer3.2.bn2.weight", "model.feature_extractor.backbone.layer3.2.bn2.bias", "model.feature_extractor.backbone.layer3.2.bn2.running_mean", "model.feature_extractor.backbone.layer3.2.bn2.running_var", "model.feature_extractor.backbone.layer3.2.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer3.2.conv3.weight", "model.feature_extractor.backbone.layer3.2.bn3.weight", "model.feature_extractor.backbone.layer3.2.bn3.bias", "model.feature_extractor.backbone.layer3.2.bn3.running_mean", "model.feature_extractor.backbone.layer3.2.bn3.running_var", "model.feature_extractor.backbone.layer3.2.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer3.3.conv1.weight", "model.feature_extractor.backbone.layer3.3.bn1.weight", "model.feature_extractor.backbone.layer3.3.bn1.bias", "model.feature_extractor.backbone.layer3.3.bn1.running_mean", "model.feature_extractor.backbone.layer3.3.bn1.running_var", "model.feature_extractor.backbone.layer3.3.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer3.3.conv2.weight", "model.feature_extractor.backbone.layer3.3.bn2.weight", "model.feature_extractor.backbone.layer3.3.bn2.bias", "model.feature_extractor.backbone.layer3.3.bn2.running_mean", "model.feature_extractor.backbone.layer3.3.bn2.running_var", "model.feature_extractor.backbone.layer3.3.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer3.3.conv3.weight", "model.feature_extractor.backbone.layer3.3.bn3.weight", "model.feature_extractor.backbone.layer3.3.bn3.bias", "model.feature_extractor.backbone.layer3.3.bn3.running_mean", "model.feature_extractor.backbone.layer3.3.bn3.running_var", "model.feature_extractor.backbone.layer3.3.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer3.4.conv1.weight", "model.feature_extractor.backbone.layer3.4.bn1.weight", "model.feature_extractor.backbone.layer3.4.bn1.bias", "model.feature_extractor.backbone.layer3.4.bn1.running_mean", "model.feature_extractor.backbone.layer3.4.bn1.running_var", "model.feature_extractor.backbone.layer3.4.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer3.4.conv2.weight", "model.feature_extractor.backbone.layer3.4.bn2.weight", "model.feature_extractor.backbone.layer3.4.bn2.bias", "model.feature_extractor.backbone.layer3.4.bn2.running_mean", "model.feature_extractor.backbone.layer3.4.bn2.running_var", "model.feature_extractor.backbone.layer3.4.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer3.4.conv3.weight", "model.feature_extractor.backbone.layer3.4.bn3.weight", "model.feature_extractor.backbone.layer3.4.bn3.bias", "model.feature_extractor.backbone.layer3.4.bn3.running_mean", "model.feature_extractor.backbone.layer3.4.bn3.running_var", "model.feature_extractor.backbone.layer3.4.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer3.5.conv1.weight", "model.feature_extractor.backbone.layer3.5.bn1.weight", "model.feature_extractor.backbone.layer3.5.bn1.bias", "model.feature_extractor.backbone.layer3.5.bn1.running_mean", "model.feature_extractor.backbone.layer3.5.bn1.running_var", "model.feature_extractor.backbone.layer3.5.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer3.5.conv2.weight", "model.feature_extractor.backbone.layer3.5.bn2.weight", "model.feature_extractor.backbone.layer3.5.bn2.bias", "model.feature_extractor.backbone.layer3.5.bn2.running_mean", "model.feature_extractor.backbone.layer3.5.bn2.running_var", "model.feature_extractor.backbone.layer3.5.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer3.5.conv3.weight", "model.feature_extractor.backbone.layer3.5.bn3.weight", "model.feature_extractor.backbone.layer3.5.bn3.bias", "model.feature_extractor.backbone.layer3.5.bn3.running_mean", "model.feature_extractor.backbone.layer3.5.bn3.running_var", "model.feature_extractor.backbone.layer3.5.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer4.0.conv1.weight", "model.feature_extractor.backbone.layer4.0.bn1.weight", "model.feature_extractor.backbone.layer4.0.bn1.bias", "model.feature_extractor.backbone.layer4.0.bn1.running_mean", "model.feature_extractor.backbone.layer4.0.bn1.running_var", "model.feature_extractor.backbone.layer4.0.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer4.0.conv2.weight", "model.feature_extractor.backbone.layer4.0.bn2.weight", "model.feature_extractor.backbone.layer4.0.bn2.bias", "model.feature_extractor.backbone.layer4.0.bn2.running_mean", "model.feature_extractor.backbone.layer4.0.bn2.running_var", "model.feature_extractor.backbone.layer4.0.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer4.0.conv3.weight", "model.feature_extractor.backbone.layer4.0.bn3.weight", "model.feature_extractor.backbone.layer4.0.bn3.bias", "model.feature_extractor.backbone.layer4.0.bn3.running_mean", "model.feature_extractor.backbone.layer4.0.bn3.running_var", "model.feature_extractor.backbone.layer4.0.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer4.0.downsample.0.weight", "model.feature_extractor.backbone.layer4.0.downsample.1.weight", "model.feature_extractor.backbone.layer4.0.downsample.1.bias", "model.feature_extractor.backbone.layer4.0.downsample.1.running_mean", "model.feature_extractor.backbone.layer4.0.downsample.1.running_var", "model.feature_extractor.backbone.layer4.0.downsample.1.num_batches_tracked", "model.feature_extractor.backbone.layer4.1.conv1.weight", "model.feature_extractor.backbone.layer4.1.bn1.weight", "model.feature_extractor.backbone.layer4.1.bn1.bias", "model.feature_extractor.backbone.layer4.1.bn1.running_mean", "model.feature_extractor.backbone.layer4.1.bn1.running_var", "model.feature_extractor.backbone.layer4.1.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer4.1.conv2.weight", "model.feature_extractor.backbone.layer4.1.bn2.weight", "model.feature_extractor.backbone.layer4.1.bn2.bias", "model.feature_extractor.backbone.layer4.1.bn2.running_mean", "model.feature_extractor.backbone.layer4.1.bn2.running_var", "model.feature_extractor.backbone.layer4.1.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer4.1.conv3.weight", "model.feature_extractor.backbone.layer4.1.bn3.weight", "model.feature_extractor.backbone.layer4.1.bn3.bias", "model.feature_extractor.backbone.layer4.1.bn3.running_mean", "model.feature_extractor.backbone.layer4.1.bn3.running_var", "model.feature_extractor.backbone.layer4.1.bn3.num_batches_tracked", "model.feature_extractor.backbone.layer4.2.conv1.weight", "model.feature_extractor.backbone.layer4.2.bn1.weight", "model.feature_extractor.backbone.layer4.2.bn1.bias", "model.feature_extractor.backbone.layer4.2.bn1.running_mean", "model.feature_extractor.backbone.layer4.2.bn1.running_var", "model.feature_extractor.backbone.layer4.2.bn1.num_batches_tracked", "model.feature_extractor.backbone.layer4.2.conv2.weight", "model.feature_extractor.backbone.layer4.2.bn2.weight", "model.feature_extractor.backbone.layer4.2.bn2.bias", "model.feature_extractor.backbone.layer4.2.bn2.running_mean", "model.feature_extractor.backbone.layer4.2.bn2.running_var", "model.feature_extractor.backbone.layer4.2.bn2.num_batches_tracked", "model.feature_extractor.backbone.layer4.2.conv3.weight", "model.feature_extractor.backbone.layer4.2.bn3.weight", "model.feature_extractor.backbone.layer4.2.bn3.bias", "model.feature_extractor.backbone.layer4.2.bn3.running_mean", "model.feature_extractor.backbone.layer4.2.bn3.running_var", "model.feature_extractor.backbone.layer4.2.bn3.num_batches_tracked", "model.feature_extractor.backbone.fc.weight", "model.feature_extractor.backbone.fc.bias". 

And this is the code I'm using:

        config = get_configurable_parameters(config_path=my_args['config'])
        config.trainer.resume_from_checkpoint = str(my_args['weights'])
        config.visualization.show_images = my_args['show']
        config.visualization.mode = my_args['visualization_mode']
        if my_args['output']:  # overwrite save path
            config.visualization.save_images = True
            config.visualization.image_save_path = my_args['output']
        else:
            config.visualization.save_images = False

        model = get_model(config)
        callbacks = get_callbacks(config)

        trainer = Trainer(callbacks=callbacks, **config.trainer)

        transform_config = config.dataset.transform_config.val if "transform_config" in config.dataset.keys() else None
        dataset = InferenceDataset(
            my_args['input'], image_size=tuple(config.dataset.image_size), transform_config=transform_config
        )
        dataloader = DataLoader(dataset)
        trainer.predict(model=model, dataloaders=[dataloader])

With weights trained on Colab, and doing inference on CPU. This is the models/patchcore/config.yaml file, almost standard:

dataset:
  name: mvtec #options: [mvtec, btech, folder]
  format: mvtec
  path: ./datasets/Tesmec
  task: segmentation
  category: rocks
  image_size: 448
  train_batch_size: 32
  test_batch_size: 1
  num_workers: 8
  transform_config:
    train: null
    val: null
  create_validation_set: false
  tiling:
    apply: false
    tile_size: 64
    stride: null
    remove_border_count: 0
    use_random_tiling: False
    random_tile_count: 16

model:
  name: patchcore
  backbone: wide_resnet50_2
  pre_trained: true
  layers:
    - layer1
    - layer2
    - layer3
  coreset_sampling_ratio: 0.1
  num_neighbors: 9
  normalization_method: min_max # options: [null, min_max, cdf]

metrics:
  image:
    - F1Score
    - AUROC
  pixel:
    - F1Score
    - AUROC
  threshold:
    image_default: 0
    pixel_default: 0
    adaptive: true

visualization:
  show_images: False # show images on the screen
  save_images: True # save images to the file system
  log_images: True # log images to the available loggers (if any)
  image_save_path: null # path to which images will be saved
  mode: full # options: ["full", "simple"]

project:
  seed: 42
  path: ./results

logging:
  logger: [] # options: [tensorboard, wandb, csv] or combinations.
  log_graph: false # Logs the model graph to respective logger.

optimization:
  export_mode: null # options: onnx, openvino

# PL Trainer Args. Don't add extra parameter here.
trainer:
  accelerator: auto # <"cpu", "gpu", "tpu", "ipu", "hpu", "auto">
  accumulate_grad_batches: 1
  amp_backend: native
  auto_lr_find: false
  auto_scale_batch_size: false
  auto_select_gpus: false
  benchmark: false
  check_val_every_n_epoch: 1 # Don't validate before extracting features.
  default_root_dir: null
  detect_anomaly: false
  deterministic: false
  devices: 1
  enable_checkpointing: true
  enable_model_summary: true
  enable_progress_bar: true
  fast_dev_run: false
  gpus: null # Set automatically
  gradient_clip_val: 0
  ipus: null
  limit_predict_batches: 1.0
  limit_test_batches: 1.0
  limit_train_batches: 1.0
  limit_val_batches: 1.0
  log_every_n_steps: 50
  log_gpu_memory: null
  max_epochs: 1
  max_steps: -1
  max_time: null
  min_epochs: null
  min_steps: null
  move_metrics_to_cpu: false
  multiple_trainloader_mode: max_size_cycle
  num_nodes: 1
  num_processes: null
  num_sanity_val_steps: 0
  overfit_batches: 0.0
  plugins: null
  precision: 32
  profiler: null
  reload_dataloaders_every_n_epochs: 0
  replace_sampler_ddp: true
  strategy: null
  sync_batchnorm: false
  tpu_cores: null
  track_grad_norm: -1
  val_check_interval: 1.0 # Don't validate before extracting features.

[alexriedel1]

@alevangel you might have not installed the same anomalib version on colab and locally where your are running your inference

[alevangel]

@alexriedel1 on Colab I'm doing:
!git clone https://github.com/openvinotoolkit/anomalib

While on local I'm on the last commit of the main branch, SHA:
bd369190cdb49f22a22ebd058ea4af46f50aee26

Those versions are not the same?

[alexriedel1]

On loading the state dictionary, it looks like the saved model file was trained before using timm models for feature extraction:
Unexpected key(s) in state_dict: "model.feature_extractor.backbone.layer1.0.bn2.running_var"

anomalib/anomalib/models/components/feature_extractors/feature_extractor.py

Line 39 in e4f8499

self.backbone = backbone

Where as the model defined for loading the weights is the most recent one using the timm feature extractor:
Missing key(s) in state_dict: "model.feature_extractor.feature_extractor.layer1.0.bn2.running_var"

anomalib/anomalib/models/components/feature_extractors/feature_extractor.py

Lines 43 to 49 in bd36919

	self.feature_extractor = timm.create_model(
	backbone,
	pretrained=pre_trained,
	features_only=True,
	exportable=True,
	out_indices=self.idx,
	)

Are you sure you trained the model with the most recent version of anomalib?

[alevangel]

@alexriedel1 I'm quite sure that I'm on the same commit on both sides:
!git checkout bd369190cdb49f22a22ebd058ea4af46f50aee26

And I'm training on colab with the method:
!python tools/train.py --model "patchcore"

Do you have weights for leather/patchcore trained with the most recent version? So that maybe I can try on my local to see if the problem persists.

[alexriedel1]

Do you have weights for leather/patchcore trained with the most recent version? So that maybe I can try on my local.

Unfortunately I don't. Can you show the code you are using for loading the trained model?

[alevangel]

This is the full code that I'm using in local:

from pytorch_lightning import Trainer
from torch.utils.data import DataLoader

from anomalib.config import get_configurable_parameters
from anomalib.data.inference import InferenceDataset
from anomalib.models import get_model
from anomalib.utils.callbacks import get_callbacks

        my_args = {'config': 'anomalib/models/patchcore/config.yaml',
                   'weights': f'results/patchcore/mvtec/leather/weights/model.ckpt',
                   'input': file_path,
                   'output': 'app/static/results/',
                   'visualization_mode': 'simple',
                   'show': False}

        config = get_configurable_parameters(config_path=my_args['config'])
        config.trainer.resume_from_checkpoint = str(my_args['weights'])
        config.visualization.show_images = my_args['show']
        config.visualization.mode = my_args['visualization_mode']
        if my_args['output']:  # overwrite save path
            config.visualization.save_images = True
            config.visualization.image_save_path = my_args['output']
        else:
            config.visualization.save_images = False

        model = get_model(config)
        callbacks = get_callbacks(config)

        trainer = Trainer(callbacks=callbacks, **config.trainer)

        transform_config = config.dataset.transform_config.val if "transform_config" in config.dataset.keys() else None
        dataset = InferenceDataset(
            my_args['input'], image_size=tuple(config.dataset.image_size), transform_config=transform_config
        )
        dataloader = DataLoader(dataset)
        trainer.predict(model=model, dataloaders=[dataloader])

Also I had to change this line (as stated in this issue):

anomalib/anomalib/utils/callbacks/model_loader.py

Line 38 in bd36919

pl_module.load_state_dict(torch.load(self.weights_path)["state_dict"])

to:
pl_module.load_state_dict(torch.load(self.weights_path, map_location=torch.device('cpu')["state_dict"])

This is also my env:

absl-py==1.2.0
aiohttp==3.8.1
aiosignal==1.2.0
albumentations==1.2.1
analytics-python==1.4.0
-e git+https://github.com/openvinotoolkit/anomalib@bd369190cdb49f22a22ebd058ea4af46f50aee26#egg=anomalib
antlr4-python3-runtime==4.9.3
anyio==3.6.1
async-timeout==4.0.2
attrs==22.1.0
backoff==1.10.0
bcrypt==4.0.0
cachetools==5.2.0
certifi==2022.6.15
cffi==1.15.1
charset-normalizer==2.1.1
click==8.1.3
colorama==0.4.5
cryptography==37.0.4
cycler==0.11.0
docker-pycreds==0.4.0
docstring-parser==0.14.1
einops==0.4.1
fastapi==0.81.0
ffmpy==0.3.0
Flask==2.2.2
fonttools==4.37.1
frozenlist==1.3.1
fsspec==2022.8.2
gitdb==4.0.9
GitPython==3.1.27
google-auth==2.11.0
google-auth-oauthlib==0.4.6
gradio==3.2
grpcio==1.47.0
h11==0.12.0
httpcore==0.15.0
httpx==0.23.0
idna==3.3
imageio==2.21.2
imgaug==0.4.0
importlib-metadata==4.12.0
itsdangerous==2.1.2
Jinja2==3.1.2
joblib==1.1.0
jsonargparse==4.13.2
kiwisolver==1.4.4
kornia==0.6.7
linkify-it-py==1.0.3
Markdown==3.4.1
markdown-it-py==2.1.0
MarkupSafe==2.1.1
matplotlib==3.5.3
mdit-py-plugins==0.3.0
mdurl==0.1.2
monotonic==1.6
multidict==6.0.2
networkx==2.8.6
numpy==1.23.2
oauthlib==3.2.0
omegaconf==2.2.3
opencv-python==4.6.0.66
opencv-python-headless==4.6.0.66
orjson==3.8.0
packaging==21.3
pandas==1.4.4
paramiko==2.11.0
pathtools==0.1.2
Pillow==9.2.0
promise==2.3
protobuf==3.19.4
psutil==5.9.1
pyasn1==0.4.8
pyasn1-modules==0.2.8
pycparser==2.21
pycryptodome==3.15.0
pydantic==1.10.1
pyDeprecate==0.3.2
pydub==0.25.1
PyNaCl==1.5.0
pyparsing==3.0.9
python-dateutil==2.8.2
python-multipart==0.0.5
pytorch-lightning==1.6.5
pytz==2022.2.1
PyWavelets==1.3.0
PyYAML==6.0
qudida==0.0.4
requests==2.28.1
requests-oauthlib==1.3.1
rfc3986==1.5.0
rsa==4.9
scikit-image==0.19.3
scikit-learn==1.1.2
scipy==1.9.1
sentry-sdk==1.9.6
setproctitle==1.3.2
Shapely==1.8.4
shortuuid==1.0.9
six==1.16.0
smmap==5.0.0
sniffio==1.2.0
starlette==0.19.1
tensorboard==2.10.0
tensorboard-data-server==0.6.1
tensorboard-plugin-wit==1.8.1
threadpoolctl==3.1.0
tifffile==2022.8.12
timm==0.5.4
torch==1.11.0
torchmetrics==0.9.1
torchtext==0.12.0
torchvision==0.12.0
tqdm==4.64.0
typing_extensions==4.3.0
uc-micro-py==1.0.1
urllib3==1.26.12
uvicorn==0.18.3
wandb==0.12.17
websockets==10.3
Werkzeug==2.2.2
yarl==1.8.1
zipp==3.8.1

[alevangel]

@alexriedel1 actually I think as you that is a version problem, running inferences on Colab seems work:

!python tools/inference/lightning_inference.py \
    --config anomalib/models/patchcore/config.yaml \
    --weights results/patchcore/mvtec/leather/weights/model.ckpt \
    --input datasets/MVTec/leather/test/cut/000.png \
    --output test_results

But I am pretty sure that is the same commit on both side... where is the trick?! :'(

[alexriedel1]

Hm. can you please show the output of

import anomalib
print(anomalib.__version__) 

on both colab and your local machine?

[alevangel]

@alexriedel1 on both sides is
__version__ = "0.3.5"

[alevangel]

@alexriedel1 oh gosh! It was a messy with the colab versions, I found the problem where was... In the big block of isntalling libraries I also installed anomalib (while also cloning the git repo), and this created an overlapping in the versions.

Now it works :') Thanks!!