# qixinyi1122-lucky/did architecture

> Architecture diagram of the GitHub repository qixinyi1122-lucky/did, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-11-18). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/qixinyi1122-lucky/did
- Repository: https://github.com/qixinyi1122-lucky/did

## Overview

DID detects AI-generated images by reconstructing inputs with guided diffusion, deriving first- and second-order residual features, training detectors on those features, and evaluating a two-branch fusion. The graph preserves the documented reconstruction outputs and the detector/fusion stages, plus the shared training and evaluation layers. Edges between sampled modules are supported by source excerpts or the SOURCE INDEX; residual and fusion relationships come from the README. Some runtime details, including detector inference, are unsampled, so uncertain links are omitted.

Main entry points: [`reconstruct/recon_metrics.py`](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/recon_metrics.py) (Feature extraction), [`detector/train.py`](https://github.com/qixinyi1122-lucky/did/blob/main/detector/train.py) (Detector training), [`fusion/evaluate.py`](https://github.com/qixinyi1122-lucky/did/blob/main/fusion/evaluate.py) (Two-branch fusion), [`reconstruct/guided_diffusion/script_util.py`](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/script_util.py) (Model factory)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_reconstruction["Reconstruction"]
  node_recon_cli["Feature extraction<br/>[recon_metrics.py]"]
  node_diffusion_factory["Model factory<br/>[script_util.py]"]
  node_diffusion_process["Diffusion sampler"]
  node_diffusion_schedule["Sampling schedules<br/>[respace.py]"]
  node_unet["Diffusion U-Net<br/>[unet.py]"]
  node_diffusion_nn["Diffusion NN utilities<br/>[nn.py]"]
  node_diffusion_runtime["Diffusion runtime<br/>[dist_util.py]"]
  node_diffusion_data["Image loader<br/>[image_datasets.py]"]
  node_diffusion_logger["Diffusion logging<br/>[logger.py]"]
  node_diffusion_fp16["Mixed precision<br/>[fp16_util.py]"]
  node_diffusion_resampling["Timestep resampling<br/>[resample.py]"]
  node_diffusion_losses["Diffusion losses<br/>[losses.py]"]
  node_feature_roots[("Residual feature roots")]
end

subgraph group_detector["Detector"]
  node_train_entry["Detector training<br/>[train.py]"]
  node_detector_test["Detector testing<br/>[test.py]"]
end

subgraph group_shared["Shared ML"]
  node_dataset_loader["Feature datasets<br/>[datasets.py]"]
  node_config["Experiment config<br/>[config.py]"]
  node_trainer["Model trainer<br/>[trainer.py]"]
  node_resnet["ResNet classifier<br/>[resnet.py]"]
  node_training_utils["Training utilities<br/>[utils.py]"]
  node_early_stopping["Early stopping<br/>[earlystop.py]"]
  node_warmup["LR warmup<br/>[warmup.py]"]
  node_validation["Detector validation<br/>[eval.py]"]
end

subgraph group_fusion["Fusion"]
  node_fusion_eval["Two-branch fusion<br/>[evaluate.py]"]
  node_dire_root[("DIRE feature root")]
  node_diff_root[("DIFF feature root")]
  node_fusion_result["Detection result"]
end

node_image_owner(("Dataset provider"))

node_image_owner -->|"provides images"| node_recon_cli
node_recon_cli -->|"sets up device"| node_diffusion_runtime
node_recon_cli -->|"configures logging"| node_diffusion_logger
node_recon_cli -->|"loads images"| node_diffusion_data
node_recon_cli -->|"creates model"| node_diffusion_factory
node_diffusion_factory -->|"builds diffusion"| node_diffusion_process
node_diffusion_factory -->|"builds schedule"| node_diffusion_schedule
node_diffusion_factory -->|"builds U-Net"| node_unet
node_unet -->|"uses NN utilities"| node_diffusion_nn
node_recon_cli -->|"writes residual maps"| node_feature_roots
node_feature_roots -->|"supplies features"| node_train_entry
node_train_entry -->|"reads settings"| node_config
node_train_entry -->|"creates loader"| node_dataset_loader
node_train_entry -->|"trains model"| node_trainer
node_trainer -->|"loads network"| node_training_utils
node_training_utils -->|"constructs classifier"| node_resnet
node_train_entry -->|"validates model"| node_validation
node_validation -->|"creates loader"| node_dataset_loader
node_train_entry -->|"checks stopping"| node_early_stopping
node_trainer -->|"uses scheduler"| node_warmup
node_dire_root -->|"provides first-order features"| node_fusion_eval
node_diff_root -->|"provides second-order features"| node_fusion_eval
node_fusion_eval -->|"reads settings"| node_config
node_fusion_eval -->|"uses evaluation"| node_validation
node_fusion_eval -->|"combines decisions"| node_fusion_result

click node_recon_cli "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/recon_metrics.py"
click node_diffusion_factory "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/script_util.py"
click node_diffusion_process "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/gaussian_diffusion.py"
click node_diffusion_schedule "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/respace.py"
click node_unet "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/unet.py"
click node_diffusion_nn "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/nn.py"
click node_diffusion_runtime "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/dist_util.py"
click node_diffusion_data "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/image_datasets.py"
click node_diffusion_logger "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/logger.py"
click node_diffusion_fp16 "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/fp16_util.py"
click node_diffusion_resampling "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/resample.py"
click node_diffusion_losses "https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/losses.py"
click node_train_entry "https://github.com/qixinyi1122-lucky/did/blob/main/detector/train.py"
click node_detector_test "https://github.com/qixinyi1122-lucky/did/blob/main/detector/test.py"
click node_dataset_loader "https://github.com/qixinyi1122-lucky/did/blob/main/common/datasets.py"
click node_config "https://github.com/qixinyi1122-lucky/did/blob/main/common/config.py"
click node_trainer "https://github.com/qixinyi1122-lucky/did/blob/main/common/trainer.py"
click node_resnet "https://github.com/qixinyi1122-lucky/did/blob/main/common/networks/resnet.py"
click node_training_utils "https://github.com/qixinyi1122-lucky/did/blob/main/common/utils.py"
click node_early_stopping "https://github.com/qixinyi1122-lucky/did/blob/main/common/earlystop.py"
click node_warmup "https://github.com/qixinyi1122-lucky/did/blob/main/common/warmup.py"
click node_validation "https://github.com/qixinyi1122-lucky/did/blob/main/common/eval.py"
click node_fusion_eval "https://github.com/qixinyi1122-lucky/did/blob/main/fusion/evaluate.py"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_recon_cli,node_diffusion_factory,node_diffusion_process,node_diffusion_schedule,node_unet,node_diffusion_nn,node_diffusion_runtime,node_diffusion_data,node_diffusion_logger,node_diffusion_fp16,node_diffusion_resampling,node_diffusion_losses,node_feature_roots toneBlue
class node_train_entry,node_detector_test toneAmber
class node_dataset_loader,node_config,node_trainer,node_resnet,node_training_utils,node_early_stopping,node_warmup,node_validation toneMint
class node_fusion_eval,node_dire_root,node_diff_root,node_fusion_result toneRose
class node_image_owner toneIndigo
```

## Components

### Reconstruction

- [Feature extraction](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/recon_metrics.py): `reconstruct/recon_metrics.py`
- [Model factory](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/script_util.py): `reconstruct/guided_diffusion/script_util.py`
- [Diffusion sampler](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/gaussian_diffusion.py): `reconstruct/guided_diffusion/gaussian_diffusion.py`
- [Sampling schedules](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/respace.py): `reconstruct/guided_diffusion/respace.py`
- [Diffusion U-Net](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/unet.py): `reconstruct/guided_diffusion/unet.py`
- [Diffusion NN utilities](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/nn.py): `reconstruct/guided_diffusion/nn.py`
- [Diffusion runtime](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/dist_util.py): `reconstruct/guided_diffusion/dist_util.py`
- [Image loader](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/image_datasets.py): `reconstruct/guided_diffusion/image_datasets.py`
- [Diffusion logging](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/logger.py): `reconstruct/guided_diffusion/logger.py`
- [Mixed precision](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/fp16_util.py): `reconstruct/guided_diffusion/fp16_util.py`
- [Timestep resampling](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/resample.py): `reconstruct/guided_diffusion/resample.py`
- [Diffusion losses](https://github.com/qixinyi1122-lucky/did/blob/main/reconstruct/guided_diffusion/losses.py): `reconstruct/guided_diffusion/losses.py`
- **Residual feature roots**

### Detector

- [Detector training](https://github.com/qixinyi1122-lucky/did/blob/main/detector/train.py): `detector/train.py`
- [Detector testing](https://github.com/qixinyi1122-lucky/did/blob/main/detector/test.py): `detector/test.py`

### Shared ML

- [Feature datasets](https://github.com/qixinyi1122-lucky/did/blob/main/common/datasets.py): `common/datasets.py`
- [Experiment config](https://github.com/qixinyi1122-lucky/did/blob/main/common/config.py): `common/config.py`
- [Model trainer](https://github.com/qixinyi1122-lucky/did/blob/main/common/trainer.py): `common/trainer.py`
- [ResNet classifier](https://github.com/qixinyi1122-lucky/did/blob/main/common/networks/resnet.py): `common/networks/resnet.py`
- [Training utilities](https://github.com/qixinyi1122-lucky/did/blob/main/common/utils.py): `common/utils.py`
- [Early stopping](https://github.com/qixinyi1122-lucky/did/blob/main/common/earlystop.py): `common/earlystop.py`
- [LR warmup](https://github.com/qixinyi1122-lucky/did/blob/main/common/warmup.py): `common/warmup.py`
- [Detector validation](https://github.com/qixinyi1122-lucky/did/blob/main/common/eval.py): `common/eval.py`

### Fusion

- [Two-branch fusion](https://github.com/qixinyi1122-lucky/did/blob/main/fusion/evaluate.py): `fusion/evaluate.py`
- **DIRE feature root**
- **DIFF feature root**
- **Detection result**

### Other components

- **Dataset provider**

## Connections

- Dataset provider → Feature extraction: provides images (evidence: `README.md`)
- Feature extraction → Diffusion runtime: sets up device (evidence: `reconstruct/recon_metrics.py`)
- Feature extraction → Diffusion logging: configures logging (evidence: `reconstruct/recon_metrics.py`)
- Feature extraction → Image loader: loads images (evidence: `reconstruct/recon_metrics.py`)
- Feature extraction → Model factory: creates model (evidence: `reconstruct/recon_metrics.py`)
- Model factory → Diffusion sampler: builds diffusion (evidence: `reconstruct/guided_diffusion/script_util.py`)
- Model factory → Sampling schedules: builds schedule (evidence: `reconstruct/guided_diffusion/script_util.py`)
- Model factory → Diffusion U-Net: builds U-Net (evidence: `reconstruct/guided_diffusion/script_util.py`)
- Diffusion U-Net → Diffusion NN utilities: uses NN utilities (evidence: `reconstruct/guided_diffusion/unet.py`)
- Feature extraction → Residual feature roots: writes residual maps (evidence: `README.md`)
- Residual feature roots → Detector training: supplies features (evidence: `README.md`)
- Detector training → Experiment config: reads settings (evidence: `detector/train.py`)
- Detector training → Feature datasets: creates loader (evidence: `detector/train.py`)
- Detector training → Model trainer: trains model (evidence: `detector/train.py`)
- Model trainer → Training utilities: loads network (evidence: `common/trainer.py`)
- Training utilities → ResNet classifier: constructs classifier (evidence: `common/utils.py`)
- Detector training → Detector validation: validates model (evidence: `detector/train.py`)
- Detector validation → Feature datasets: creates loader (evidence: `common/eval.py`)
- Detector training → Early stopping: checks stopping (evidence: `detector/train.py`)
- Model trainer → LR warmup: uses scheduler (evidence: `common/trainer.py`)
- DIRE feature root → Two-branch fusion: provides first-order features (evidence: `README.md`)
- DIFF feature root → Two-branch fusion: provides second-order features (evidence: `README.md`)
- Two-branch fusion → Experiment config: reads settings (evidence: `fusion/evaluate.py`)
- Two-branch fusion → Detector validation: uses evaluation (evidence: `fusion/evaluate.py`)
- Two-branch fusion → Detection result: combines decisions (evidence: `README.md`)
