[Bug]: GANomaly prediction score not clear

Describe the bug

Hello,

First of all, thank you for the great repository and all the work that was put into it.

I've been using the GANomaly model on a custom dataset (sixray) and I'm having a hard time trying to analyse results through visualization.

As far as I understood, there are two options to set output scores.

Without normalization The model may predict scores bigger than 1 without normalization min_max. However I get always "abnormal" labels on my results. Probably because all of the images (normal and abnormal) are bigger that the adaptive threshold. Something that doens't make sense to me.
With normalization If I use normalization "min_max" the threshold becomes automatically 0.5 and the "adaptive_threshold" doens't do anything anymore. With this method I am getting F1 Score = 0 and all of the output images display "abnormal" labels. The "normal" have in fact lower scores (near 50%) but are stil above 0.5.

Dataset

Other (please specify in the text field below)

Model

GANomaly

Steps to reproduce the behavior

OS information

OS information:

OS: [e.g. Ubuntu 20.04]
Python version: [e.g. 3.8.10]
Anomalib version: [e.g. 0.3.6]
PyTorch version: [e.g. 1.9.0]
CUDA/cuDNN version: [e.g. 11.1]
GPU models and configuration: [e.g. 2x GeForce RTX 3090]
Any other relevant information: [e.g. I'm using a custom dataset]

Expected behavior

Ideally I would like to see in the image a result between 0 ad 1 (0% and 100%) but with an adaptive threshold as well. If the image was considered nomral it would appear "Normal: 'score lower than threshold'. If the image was in fact considered anomalous it would appear "Anomalous": 'score lower than threshold'.

In the article "GANomaly: Semi-Supervised Anomaly Detection via Adversarial Training" on image 6a, the results are between 0 and 1 and the threshold is not 0.5.

Screenshots

No response

Pip/GitHub

pip

What version/branch did you use?

No response

Configuration YAML

dataset:
  name: sixray
  format: folder
  path: ./datasets/sixray
  normal_dir: normal
  abnormal_dir: abnormal
  task: classification
  mask_dir: null
  normal_test_dir: real
  extensions: null
  split_ratio: 0.2  # normal images ratio to create a test split
  seed: 0
  train_batch_size: 32
  eval_batch_size: 32
  inference_batch_size: 32
  num_workers: 8
  image_size: 256 # dimensions to which images are resized (mandatory)
  center_crop: null # dimensions to which images are center-cropped after resizing (optional)
  normalization: imagenet # data distribution to which the images will be normalized: [none, imagenet]
  transform_config:
    train: null
    eval: null
  test_split_mode: from_dir # options: [from_dir, synthetic]
  test_split_ratio: 0.2 # fraction of train images held out testing (usage depends on test_split_mode)
  val_split_mode: same_as_test # options: [same_as_test, from_test, synthetic]
  val_split_ratio: 0.5 # fraction of train/test images held out for validation (usage depends on val_split_mode)
  tiling:
    apply: true
    tile_size: 64
    stride: null
    remove_border_count: 0
    use_random_tiling: False
    random_tile_count: 16

model:
  name: ganomaly
  latent_vec_size: 100
  n_features: 64
  extra_layers: 0
  add_final_conv: true 
  early_stopping:
    patience: 3
    metric: image_AUROC
    mode: max
  lr: 0.0002
  beta1: 0.5
  beta2: 0.999
  wadv: 1
  wcon: 50
  wenc: 1
  normalization_method: min_max

metrics:
  image:
    - F1Score
    - AUROC
    - AUPRO
  threshold:
    method: adaptive #options: [adaptive, manual]
    manual_image: null

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: [comet, tensorboard, wandb, csv] or combinations.
  log_graph: true # Logs the model graph to respective logger.

optimization:
  export_mode: null

# PL Trainer Args. Don't add extra parameter here.
trainer:
  enable_checkpointing: true
  default_root_dir: null
  gradient_clip_val: 0
  gradient_clip_algorithm: norm
  num_nodes: 1
  devices: 1
  enable_progress_bar: true
  overfit_batches: 0.0
  track_grad_norm: -1
  check_val_every_n_epoch: 2 # Don't validate before extracting features.
  fast_dev_run: false
  accumulate_grad_batches: 1
  max_epochs: 100
  max_steps: -1
  min_steps: null
  max_time: null
  limit_train_batches: 1.0
  limit_val_batches: 1.0
  limit_test_batches: 1.0
  limit_predict_batches: 1.0
  val_check_interval: 1.0 # Don't validate before extracting features.
  log_every_n_steps: 50
  accelerator: auto # <"cpu", "gpu", "tpu", "ipu", "hpu", "auto">
  strategy: null
  sync_batchnorm: false
  precision: 32
  enable_model_summary: true
  num_sanity_val_steps: 0
  profiler: null
  benchmark: false
  deterministic: false
  reload_dataloaders_every_n_epochs: 0
  auto_lr_find: false
  replace_sampler_ddp: true
  detect_anomaly: false
  auto_scale_batch_size: false
  plugins: null
  move_metrics_to_cpu: false
  multiple_trainloader_mode: max_size_cycle

Logs

Code of Conduct

[X] I agree to follow this project's Code of Conduct

openvinotoolkit / anomalib