# yuchendoudou/tode architecture

> Architecture diagram of the GitHub repository yuchendoudou/tode, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2025-04-21). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/yuchendoudou/tode
- Repository: https://github.com/yuchendoudou/tode

## Overview

##### Project type and purpose
- This is a research‐oriented deep learning pipeline in PyTorch for transparent object depth estimation using a Transformer backbone (TODE-Trans).
- Key use cases: dataset preparation, model training, evaluation on multiple datasets (ClearGrasp, Omniverse, TransCG), checkpoint management, and sample inference.

##### File structure
- Top‑level scripts:
- train.py – orchestrates training loops via config files
- inference.py / sample_inference.py – perform one‐off inference for evaluation or demos
- test.py – runs validation/testing protocols
- Directories by function:
- configs/ – YAML files defining hyperparameters, dataset splits, training/inference modes
- datasets/ – PyTorch Dataset wrappers for each source (cleargrasp, omniverse, transcg, mixed)
- models/ – network definitions (swin_transformer.py, Tode.py)
- utils/ – builder, data preparation, loss (criterion), metrics, logging, helper functions
- checkpoints/ – pretrained model weights for inference/testing

##### README for additional insights
- External dependencies: HDF5, OpenEXR, PyTorch 1.11.0
- Datasets must be downloaded and arranged under a root directory
- Commands show how configs drive different training/test scenarios

1. Dataset Management
- ClearGrasp Dataset Loader
- Omniverse Dataset Loader
- TransCG Dataset Loader
- MixedDataset wrapper
2. Configuration Engine
- YAML config parser (default.yaml, train_*.yaml, inference.yaml)
3. Model Builder
- Swin Transformer backbone
- TODE‐Trans head (depth estimation)
4. Training Pipeline
- DataLoader → Model → Loss (criterion) → Optimizer → Scheduler
- Checkpointer (save/load weights)
5. Evaluation / Metrics
- Depth metrics (e.g., abs error, rmse) computed in utils/metrics.py
- Logger (console + file) in utils/logger.py
6. Inference / Sample Inference
- Inference script using inference.yaml
- Output visualization / depth map export
7. External Dependencies
- GPU / CUDA
- HDF5 / OpenEXR libraries

- Config Engine → Training Pipeline & Inference
- Dataset Management → Training Pipeline (DataLoader)
- Training Pipeline → Model Builder
- Model Builder → Compute Graph (forward/backward)
- Training Pipeline → Checkpointer (save)
- Training Pipeline → Evaluation (metrics, logger)
- Inference Script → Model Builder (load weights) → Dataset → Output Visualizer

c. Architectural patterns & design principles:
- Modular separation: clear boundaries between data loading, model definition, training logic, and utilities
- Config-driven execution: YAML files allow reproducible experiments without code edits
- Checkpoint abstraction: copycheckpoint.py handles weight transfers; training scripts use checkpoint API
- Extensibility: new datasets/models can be added by extending datasets/ or models/ directories

d. Key technologies/frameworks:
- PyTorch 1.11.0 (torch, torchvision)
- HDF5, OpenEXR for data formats
- Python 3.x ecosystem (numpy, yaml, tqdm, etc.)

##### Final recommendations
- Both control flow (script → module) and data flow (images → depth maps)

## Diagram (Mermaid)

```mermaid
flowchart TD
    subgraph "Data Layer"
        CG["ClearGrasp Dataset Loader"]:::data
        OM["Omniverse Dataset Loader"]:::data
        TC["TransCG Dataset Loader"]:::data
        MD["Mixed Dataset Wrapper"]:::data
    end

    subgraph "Configuration Engine"
        CFG["YAML Config Files"]:::config
    end

    subgraph "Model Builder"
        SW["Swin Transformer Backbone"]:::model
        TDH["TODE-Trans Head"]:::model
    end

    subgraph "Training Pipeline"
        TRS["train.py"]:::script
        DL["DataLoader"]:::pipeline
        LSS["Loss (criterion)"]:::pipeline
        OPT["Optimizer"]:::pipeline
        SCH["Scheduler"]:::pipeline
        CHK["Checkpointer"]:::script
    end

    subgraph "Evaluation & Metrics"
        MET["Metrics Computation"]:::eval
        LOG["Logger (console & file)"]:::eval
    end

    subgraph "Inference"
        INF["inference.py"]:::script
        SINF["sample_inference.py"]:::script
        OUT["Output Visualization"]:::infer
    end

    PTST["test.py"]:::script
    CKPTD["Checkpoints Directory"]:::data

    subgraph "Utilities"
        UBD["Builder"]:::script
        UPD["Data Preparation"]:::script
        UFN["Helper Functions"]:::script
        UCN["Constants"]:::script
    end

    %% Relationships
    CFG --> TRS
    CFG --> INF
    CFG --> SINF

    CG --> DL
    OM --> DL
    TC --> DL
    MD --> DL

    TRS --> DL
    DL --> SW
    SW --> TDH
    TDH --> LSS
    LSS --> OPT
    OPT --> SCH
    TRS --> CHK
    TRS --> MET
    TRS --> LOG

    INF --> SW
    INF --> TDH
    INF --> CKPTD
    INF --> OUT

    SINF --> SW
    PTST --> SW
    PTST --> TDH
    PTST --> MET
    PTST --> LOG

    CHK --> CKPTD
    CKPTD -.-> SW

    TRS --> UBD
    INF --> UBD
    UBD --> DL
    UBD --> SW
    TRS --> UPD
    INF --> UPD
    TRS --> UFN
    INF --> UFN
    TRS --> UCN
    INF --> UCN

    %% Click Events
    click CG "https://github.com/yuchendoudou/tode/blob/main/datasets/cleargrasp.py"
    click OM "https://github.com/yuchendoudou/tode/blob/main/datasets/omniverse_object.py"
    click TC "https://github.com/yuchendoudou/tode/blob/main/datasets/transcg.py"
    click MD "https://github.com/yuchendoudou/tode/blob/main/datasets/mixed_dataset.py"
    click CFG "https://github.com/yuchendoudou/tode/blob/main/configs/default.yaml"
    click SW "https://github.com/yuchendoudou/tode/blob/main/models/swin_transformer.py"
    click TDH "https://github.com/yuchendoudou/tode/blob/main/models/Tode.py"
    click TRS "https://github.com/yuchendoudou/tode/blob/main/train.py"
    click INF "https://github.com/yuchendoudou/tode/blob/main/inference.py"
    click SINF "https://github.com/yuchendoudou/tode/blob/main/sample_inference.py"
    click PTST "https://github.com/yuchendoudou/tode/blob/main/test.py"
    click CHK "https://github.com/yuchendoudou/tode/blob/main/copycheckpoint.py"
    click CKPTD "https://github.com/yuchendoudou/tode/blob/main/checkpoints/transcg_checkpoint.tar"
    click LSS "https://github.com/yuchendoudou/tode/blob/main/utils/criterion.py"
    click MET "https://github.com/yuchendoudou/tode/blob/main/utils/metrics.py"
    click LOG "https://github.com/yuchendoudou/tode/blob/main/utils/logger.py"
    click UPD "https://github.com/yuchendoudou/tode/blob/main/utils/data_preparation.py"
    click UBD "https://github.com/yuchendoudou/tode/blob/main/utils/builder.py"
    click UFN "https://github.com/yuchendoudou/tode/blob/main/utils/functions.py"
    click UCN "https://github.com/yuchendoudou/tode/blob/main/utils/constants.py"

    %% Styles
    classDef data fill:#e3f2fd,stroke:#90caf9
    classDef config fill:#fff3e0,stroke:#ffb74d
    classDef model fill:#f3e5f5,stroke:#ce93d8
    classDef pipeline fill:#e8f5e9,stroke:#66bb6a
    classDef eval fill:#ffebee,stroke:#e57373
    classDef infer fill:#fff8e1,stroke:#ffd54f
    classDef script fill:#cfd8dc,stroke:#b0bec5
```
