Best performing models for rock joint segmentation

See description for selection of best models in Github-linkhttps://github.com/tfha/syntetic_rock_joint_generation_for_ml/tree/master/scripts/results_analysis#3-download-best-trained-models Folder structure as shown below. Each experiment is packed into a zip folder. ├── box.zip/│ ├── finetune/│ │ ├── best_val_dice_joint/│ │ │ └── unet-finetune_box_50-20251211-1618/│ │ │ ├── best_model.pth│ │ │ └── stage1_best_model.pth│ │ └── best_quality_score/│ │ └── unet-finetune_box_10-20251211-1615/│ │ ├── best_model.pth│ │ └── stage1_best_model.pth│ └── simplemixed/│ ├── best_val_dice_joint/│ │ └── unet-simplemixed_box_90-20251211-1603/│ │ ├── best_metrics_model.pth│ │ └── final_model.pth│ └── best_quality_score/│ └── unet-simplemixed_box_100-20251211-1605/│ ├── best_metrics_model.pth│ └── final_model.pth└── [9 more experiment folders...]