# samuelschmidgall/gsvit architecture

> Architecture diagram of the GitHub repository samuelschmidgall/gsvit, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2025-01-03). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/samuelschmidgall/gsvit
- Repository: https://github.com/samuelschmidgall/gsvit

## Overview

##### Main Components

a) Data Processing Pipeline
- Video Input Processing
- Surgical Video Dataset Management
- Data Augmentation Layer

b) Model Architecture
- EfficientViT Base Architecture
- Classification Module
- Downstream Tasks Module
- Pre-training Module

c) Training Infrastructure
- Distributed Training System
- Model Checkpoint Management
- Custom MMCV Runner Components

##### Key Relationships to Highlight

**Raw Surgical Videos → Data Processing → EfficientViT Backbone → Task-Specific Heads**

##### Important Technical Details

- Connection between EfficientViT and the custom surgical adaptations
- Multi-task capability (classification, detection, instance segmentation)
- Pre-training and fine-tuning workflow
- Distributed training architecture

##### Specific Implementation Notes

- Rectangles: Processing modules
- Hexagons: Model components

##### Additional Elements

Annotations for:
- Model configuration paths
- Data augmentation strategies
- Training schedule options
- Checkpoint saving/loading mechanisms

## Diagram (Mermaid)

```mermaid
graph TB
    subgraph Input
        VI[Surgical Video Input]
        VPM["Video Processing Module"]:::dataproc
        SDM["Surgical Dataset Management"]:::dataset
        DAL["Data Augmentation Layer"]:::augment
    end

    subgraph Core_Architecture
        EVB["EfficientViT Backbone"]:::model
        VTC["Vision Transformer Components"]:::model
        FEM["Feature Extraction Modules"]:::model
    end

    subgraph Task_Specific
        CH["Classification Head"]:::task
        DH["Detection Head"]:::task
        ISC["Instance Segmentation"]:::task
    end

    subgraph Training_Infrastructure
        DTS["Distributed Training System"]:::infra
        CMS["Checkpoint Management"]:::infra
        CMR["Custom MMCV Runner"]:::infra
    end

    %% Data flow connections
    VI --> VPM
    VPM --> SDM
    SDM --> DAL
    DAL --> EVB
    EVB --> VTC
    VTC --> FEM
    
    FEM --> CH
    FEM --> DH
    FEM --> ISC
    
    CH --> DTS
    DH --> DTS
    ISC --> DTS
    
    DTS --> CMS
    CMS --> CMR

    %% Click events for component mapping
    click VPM "https://github.com/samuelschmidgall/gsvit/blob/main/dataloader_surgical.py"
    click SDM "https://github.com/samuelschmidgall/gsvit/tree/main/videos_to_download"
    click DAL "https://github.com/samuelschmidgall/gsvit/blob/main/EfficientViT/classification/data/threeaugment.py"
    click EVB "https://github.com/samuelschmidgall/gsvit/blob/main/EfficientViT/classification/model/efficientvit.py"
    click CH "https://github.com/samuelschmidgall/gsvit/blob/main/EfficientViT/classification/model/build.py"
    click DH "https://github.com/samuelschmidgall/gsvit/blob/main/EfficientViT/downstream/efficientvit.py"
    click DTS "https://github.com/samuelschmidgall/gsvit/blob/main/EfficientViT/downstream/dist_train.sh"
    click CMS "https://github.com/samuelschmidgall/gsvit/blob/main/EfficientViT/downstream/mmcv_custom/checkpoint.py"
    click CMR "https://github.com/samuelschmidgall/gsvit/tree/main/EfficientViT/downstream/mmcv_custom/runner"
    click ISC "https://github.com/samuelschmidgall/gsvit/blob/main/EfficientViT/downstream/configs/_base_/models/mask_rcnn_efficientvit_fpn.py"

    %% Styling
    classDef default fill:#f9f9f9,stroke:#333,stroke-width:2px
    classDef dataproc fill:#a8d1f0,stroke:#333
    classDef dataset fill:#b3e0ff,stroke:#333
    classDef augment fill:#99ccff,stroke:#333
    classDef model fill:#90EE90,stroke:#333
    classDef task fill:#FFB366,stroke:#333
    classDef infra fill:#E6E6E6,stroke:#333

    %% Legend
    subgraph Legend
        L1["Data Processing"]:::dataproc
        L2["Dataset Management"]:::dataset
        L3["Model Components"]:::model
        L4["Task Modules"]:::task
        L5["Infrastructure"]:::infra
    end
```
