🎯 [EPIC] Design and Implement the New AnomalibModule for v2.0

[samet-akcay]

Describe the epic

Design and implement the new AnomalibModule, which will serve as the core component of the Anomalib v2.0.0 release. This module will encapsulate all auxiliary operations (pre-processing, post-processing, evaluation, and visualization) as its main attributes, providing a unified and modular approach to anomaly detection pipelines.

For more details of the design, one could refer to the following slide deck.
New AnomalibModule with Auxiliary Operations.pptx

Goals

• Create a flexible and extensible AnomalibModule as the cornerstone of Anomalib v2.0.0
• Implement auxiliary operations as attributes of AnomalibModule for improved modularity and reusability
• Ensure compatibility with various model architectures and training paradigms
• Simplify the process of creating and customizing anomaly detection pipelines
• Improve overall performance and scalability, including multi-GPU support

Tasks

High Priority

• 📋 [TASK] Dataclass-based Input/Output in Anomalib #2257
• 📋 [TASK] Integrate Pre-processing as AnomalibModule Attribute #2366
• 📋 [TASK] Integrate Post-processing as AnomalibModule Attribute #2367
• 📋 [TASK] Integrate Evaluation as AnomalibModule Attribute #2368
• 📋 [TASK] Integrate Visualization as AnomalibModule Attribute #2369
• 📋 [TASK] Implement Core v2 AnomalibModule #2365
• 📋 [TASK] Implement Multi-GPU Training Support #2258
• 📋 [TASK] Add Documentation and Examples for AnomalibModule v2 #2370
• 📋 [TASK] Integration Testing and Performance Optimization #2371
• 🔨 Dataset Additions: CSV data, custom validation set, dataset filtering and splitting support #2239
• 📋 [TASK] Refactor Perlin Noise Augmenter #2429

Low Priority

• 📋 [TASK] Remove try_import function for v2.0 #2374