API Reference

This section provides detailed API documentation for the M3SGG project modules.

API Documentation:

• Datasets Module
  • Action Genome Dataset
  • EASG Dataset
  • Dataset Factory
• Core Library
  • Configuration
  • Detectors
  • Evaluation
  • Training
• Models
  • STTran Model
  • SceneLLM Model
  • Tempura Model
  • STKET Model
  • OED Model
  • VLM Model
  • Transformer Components
• CLI Module
  • Main CLI Interface
  • Training Commands
  • CLI Package
• Language Module
  • Summarization
  • Evaluation
  • Language Modeling
• Utils Module
  • Core Utilities
  • FPN Utilities
  • Visualization
  • Optimization
  • Tracking

Core Components

The M3SGG project consists of several key components:

• Datasets Module: Handles dataset loading and preprocessing

• Core Library: Configuration, models, detectors, evaluation, and training

• Models: Implementations including STTran, SceneLLM, Tempura, STKET, OED, VLM

• CLI Module: Command-line interface and utilities

• Language Module: Summarization and evaluation tools

• Utils Module: Helper functions and utilities