# plusgood-steven/id-blau architecture

> Architecture diagram of the GitHub repository plusgood-steven/id-blau, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2025-05-29). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/plusgood-steven/id-blau
- Repository: https://github.com/plusgood-steven/id-blau

## Overview

##### Project type and purpose
- ID-Blau is a research-grade image-deblurring pipeline that augments sharp images via an “implicit diffusion-based reBLurring” module, then uses existing state-of-the-art deblurring models (MIMO-UNet, Restormer, Stripformer, FFTformer) to train and evaluate performance.

##### File structure
- Top-level “PrepareCondition” directory: extracts pixel-wise blur condition maps using RAFT.
- Root scripts:
– diffusion_train.py / diffusion_inference.py → trains and runs the ID-Blau diffusion reblurring model (in models/diffusion_*.py).
– dataloader.py, dataset/ → data I/O for both reblurring and deblurring stages.
– eval_realblur.py, eval_GoPro_HIDE.m → evaluation modules.
- Subdirectories per deblurring model (FFTformer, MIMO_UNet, Restormer, Stripformer): each contains training, inference, loss, and model definition files.
- utils/: shared utilities (flow visualization, conditioning, general I/O).
- weights/: pretrained RAFT and ID-Blau checkpoint, plus deblurring model weights.

##### Incorporate README insights
- External dependencies: GoPro datasets (GOPRO_Large_all, GOPRO_Large), RAFT optical-flow model.
- Two major workflows:
(a) Data preparation & reblurring augmentation (PrepareCondition → diffusion module).
(b) Deblurring model training, inference, and evaluation (per-model scripts + eval scripts).

##### Main components & interactions
a. Data Storage & Source
– Raw datasets: GOPRO_Large_all (for flow), GOPRO_Large (sharp images), plus optional RealBlur/HIDE test sets.
– Stored under dataset/ and PrepareCondition/demo-frames.
b. Condition Extraction Module (PrepareCondition)
– Input: GOPRO_Large_all frames + RAFT weights
– Output: blur_condition_maps saved under dataset/GOPRO_flow or custom path
c. Reblurring/Diffusion Module
– diffusion_train.py uses models/diffusion_network.py to learn mapping (sharp + condition → blurred)
– diffusion_inference.py with ID_Blau.pth generates augmented blurred dataset (dataset/GOPRO_Large_Reblur)
d. Deblurring Models Suite
– Four parallel model pipelines: FFTformer, MIMO_UNet, Restormer, Stripformer
– Each has its own train scripts (e.g., deblur_train[_pretrained][_realblur].py), inference script (deblur_predict.py), and model definitions (model.py or layers.py)
– Shared loss utilities under each folder or in root losses/ and loss_util.py
e. Evaluation Module
– MATLAB script eval_GoPro_HIDE.m for GoPro/HIDE metrics
– Python script eval_realblur.py for RealBlur_J / RealBlur_R
f. Utilities & Common Code
– dataloader.py, utils/*.py for data loaders, augmentations, flow visualization, configuration

##### Architectural patterns & design principles
- Modular pipelines: clear separation between data prep (PrepareCondition), augmentation (Diffusion), deblurring (per-model folders), and evaluation
- Shared components: utility functions and data loaders are reused across modules

##### Essential elements to capture
- Precise component labels (e.g., “PrepareCondition/generate_condition.py,” “diffusion_train.py,” “MIMO_UNet/deblur_predict.py”)
- Data flow direction and dependencies (e.g., diffusion inference requires pretrained ID_Blau.pth)
- Distinguish code modules vs. data vs. external models/services

## Diagram (Mermaid)

```mermaid
flowchart TD
    subgraph "Data Storage & Source"
        DS1["dataset/"]:::data
        DS2["PrepareCondition/demo-frames"]:::data
        DS3["weights/"]:::data
    end

    DS1 -->|"raw frames"| PC_Module["PrepareCondition Module"]:::compute

    subgraph "Condition Extraction (PrepareCondition)"
        PCscript["generate_condition.py"]:::compute
        RAFT["raft.py"]:::compute
        extractor["extractor.py"]:::compute
        datasets_util["datasets.py"]:::compute
        flow_viz["flow_viz.py"]:::compute
        frame_utils["frame_utils.py"]:::compute
        train["train.py"]:::compute
        train_std["train_standard.sh"]:::compute
        train_mix["train_mixed.sh"]:::compute
    end

    PC_Module --> PCscript
    PCscript --> RAFT
    PCscript --> extractor
    extractor --> datasets_util
    RAFT --> flow_viz
    flow_viz --> DS2
    PCscript -->|"blur condition maps"| DS_condition["dataset/GOPRO_flow"]:::data

    DS_condition --> DT_Module["Diffusion Module"]:::compute

    subgraph "Reblurring / Diffusion Module"
        DT_train["diffusion_train.py"]:::compute
        DT_inf["diffusion_inference.py"]:::compute
        DN1["models/diffusion_network.py"]:::compute
        DN2["models/diffusion_model.py"]:::compute
        DLoss["models/losses.py"]:::compute
    end

    DS_condition --> DT_train
    DT_train --> DN1
    DN1 --> DN2
    DN2 --> DLoss
    DT_train -->|"trained ID_Blau.pth"| DS3
    DS3 --> DT_inf
    DT_inf -->|"augmented dataset (Reblur)"| DS_aug["dataset/GOPRO_Large_Reblur"]:::data

    DS_aug --> DM_Module["Deblurring Models Suite"]:::compute

    subgraph "Deblurring Models Suite"
        subgraph "FFTformer"
            FFT_train["deblur_train.py"]:::compute
            FFT_train_pre["deblur_train_pretrained.py"]:::compute
            FFT_train_real["deblur_train_realblur.py"]:::compute
            FFT_inf["deblur_predict.py"]:::compute
            FFT_model["model.py"]:::compute
            FFT_losses["losses.py"]:::compute
            FFT_lutil["loss_util.py"]:::compute
        end
        subgraph "MIMO-UNet"
            MIMO_train["deblur_train.py"]:::compute
            MIMO_train_pre["deblur_train_pretrained.py"]:::compute
            MIMO_train_real["deblur_train_realblur.py"]:::compute
            MIMO_inf["deblur_predict.py"]:::compute
            MIMO_model["MIMOUNet.py"]:::compute
            MIMO_layers["layers.py"]:::compute
        end
        subgraph "Restormer"
            R_train["deblur_train.py"]:::compute
            R_train_pre["deblur_train_pretrained.py"]:::compute
            R_train_real["deblur_train_realblur.py"]:::compute
            R_inf["deblur_predict.py"]:::compute
            R_model["model.py"]:::compute
            R_losses["losses.py"]:::compute
            R_lutil["loss_util.py"]:::compute
        end
        subgraph "Stripformer"
            S_train1["deblur_train_first.py"]:::compute
            S_train2["deblur_train_second.py"]:::compute
            S_train_pre["deblur_train_pretrained.py"]:::compute
            S_train_real["deblur_train_realblur.py"]:::compute
            S_inf["deblur_predict.py"]:::compute
            S_model["model.py"]:::compute
            S_losses["losses.py"]:::compute
        end
    end

    DS_aug --> FFT_train
    DS_aug --> MIMO_train
    DS_aug --> R_train
    DS_aug --> S_train1

    FFT_train --> FFT_inf
    MIMO_train --> MIMO_inf
    R_train --> R_inf
    S_train1 --> S_inf

    FFT_inf --> DP["Deblurred Output"]:::data
    MIMO_inf --> DP
    R_inf --> DP
    S_inf --> DP

    DP --> Eval_Module["Evaluation Module"]:::eval

    subgraph "Evaluation Module"
        E1["eval_GoPro_HIDE.m"]:::eval
        E2["eval_realblur.py"]:::eval
    end

    DP --> E1
    DP --> E2

    click DS1 "https://github.com/plusgood-steven/id-blau/tree/main/dataset/"
    click DS2 "https://github.com/plusgood-steven/id-blau/tree/main/PrepareCondition/demo-frames"
    click DS3 "https://github.com/plusgood-steven/id-blau/tree/main/weights/"
    click PCscript "https://github.com/plusgood-steven/id-blau/blob/main/PrepareCondition/generate_condition.py"
    click RAFT "https://github.com/plusgood-steven/id-blau/blob/main/PrepareCondition/core/raft.py"
    click extractor "https://github.com/plusgood-steven/id-blau/blob/main/PrepareCondition/core/extractor.py"
    click datasets_util "https://github.com/plusgood-steven/id-blau/blob/main/PrepareCondition/core/datasets.py"
    click flow_viz "https://github.com/plusgood-steven/id-blau/blob/main/PrepareCondition/core/utils/flow_viz.py"
    click frame_utils "https://github.com/plusgood-steven/id-blau/blob/main/PrepareCondition/core/utils/frame_utils.py"
    click train "https://github.com/plusgood-steven/id-blau/blob/main/PrepareCondition/train.py"
    click train_std "https://github.com/plusgood-steven/id-blau/blob/main/PrepareCondition/train_standard.sh"
    click train_mix "https://github.com/plusgood-steven/id-blau/blob/main/PrepareCondition/train_mixed.sh"
    click DT_train "https://github.com/plusgood-steven/id-blau/blob/main/diffusion_train.py"
    click DT_inf "https://github.com/plusgood-steven/id-blau/blob/main/diffusion_inference.py"
    click DN1 "https://github.com/plusgood-steven/id-blau/blob/main/models/diffusion_network.py"
    click DN2 "https://github.com/plusgood-steven/id-blau/blob/main/models/diffusion_model.py"
    click DLoss "https://github.com/plusgood-steven/id-blau/blob/main/models/losses.py"
    click DS_aug "https://github.com/plusgood-steven/id-blau/tree/main/dataset/GOPRO_Large_Reblur"
    click FFT_train "https://github.com/plusgood-steven/id-blau/blob/main/FFTformer/deblur_train.py"
    click FFT_train_pre "https://github.com/plusgood-steven/id-blau/blob/main/FFTformer/deblur_train_pretrained.py"
    click FFT_train_real "https://github.com/plusgood-steven/id-blau/blob/main/FFTformer/deblur_train_realblur.py"
    click FFT_inf "https://github.com/plusgood-steven/id-blau/blob/main/FFTformer/deblur_predict.py"
    click FFT_model "https://github.com/plusgood-steven/id-blau/blob/main/FFTformer/model.py"
    click FFT_losses "https://github.com/plusgood-steven/id-blau/blob/main/FFTformer/losses.py"
    click FFT_lutil "https://github.com/plusgood-steven/id-blau/blob/main/FFTformer/loss_util.py"
    click MIMO_train "https://github.com/plusgood-steven/id-blau/blob/main/MIMO_UNet/deblur_train.py"
    click MIMO_train_pre "https://github.com/plusgood-steven/id-blau/blob/main/MIMO_UNet/deblur_train_pretrained.py"
    click MIMO_train_real "https://github.com/plusgood-steven/id-blau/blob/main/MIMO_UNet/deblur_train_realblur.py"
    click MIMO_inf "https://github.com/plusgood-steven/id-blau/blob/main/MIMO_UNet/deblur_predict.py"
    click MIMO_model "https://github.com/plusgood-steven/id-blau/blob/main/MIMO_UNet/MIMOUNet.py"
    click MIMO_layers "https://github.com/plusgood-steven/id-blau/blob/main/MIMO_UNet/layers.py"
    click R_train "https://github.com/plusgood-steven/id-blau/blob/main/Restormer/deblur_train.py"
    click R_train_pre "https://github.com/plusgood-steven/id-blau/blob/main/Restormer/deblur_train_pretrained.py"
    click R_train_real "https://github.com/plusgood-steven/id-blau/blob/main/Restormer/deblur_train_realblur.py"
    click R_inf "https://github.com/plusgood-steven/id-blau/blob/main/Restormer/deblur_predict.py"
    click R_model "https://github.com/plusgood-steven/id-blau/blob/main/Restormer/model.py"
    click R_losses "https://github.com/plusgood-steven/id-blau/blob/main/Restormer/losses.py"
    click R_lutil "https://github.com/plusgood-steven/id-blau/blob/main/Restormer/loss_util.py"
    click S_train1 "https://github.com/plusgood-steven/id-blau/blob/main/Stripformer/deblur_train_first.py"
    click S_train2 "https://github.com/plusgood-steven/id-blau/blob/main/Stripformer/deblur_train_second.py"
    click S_train_pre "https://github.com/plusgood-steven/id-blau/blob/main/Stripformer/deblur_train_pretrained.py"
    click S_train_real "https://github.com/plusgood-steven/id-blau/blob/main/Stripformer/deblur_train_realblur.py"
    click S_inf "https://github.com/plusgood-steven/id-blau/blob/main/Stripformer/deblur_predict.py"
    click S_model "https://github.com/plusgood-steven/id-blau/blob/main/Stripformer/model.py"
    click S_losses "https://github.com/plusgood-steven/id-blau/blob/main/Stripformer/losses.py"
    click E1 "https://github.com/plusgood-steven/id-blau/blob/main/eval_GoPro_HIDE.m"
    click E2 "https://github.com/plusgood-steven/id-blau/blob/main/eval_realblur.py"

    classDef data fill:#ADD8E6,stroke:#000,stroke-width:1px
    classDef compute fill:#90EE90,stroke:#000,stroke-width:1px
    classDef eval fill:#FFA500,stroke:#000,stroke-width:1px
```
