language	library_name	license_name	compatibility
en	nnll	MPL-2.0 + Commons Clause 1.0	macos windows linux

negate
_{entrypoint synthetic image classifier}

A tool, research, and training library to detect the origin of images.

Quick Start

^{Terminal Powershell sh}

1.

2. uv tool install 'negate @ git+https://github.com/darkshapes/negate'

3. negate infer image.webp

Result	Translated Source
SYN	Synthetic/AI
GNE	Genuine/Human
?	High Uncertainty

Tip

To run without installing, use the uv command uvx --from 'negate @ git+https://github.com/darkshapes/negate' infer image.webp

Training

Train a new model with the following command:

negate train

Tip

type a path to an image file or directory of image files to add genuine human origin assets to the dataset add synthetic images using -s before a path

Technical Details & Research Results

Expand

Structure

Directories are located within $HOME\.local\bin\uv\tools or .local/bin/uv/tools

Data	Location	source
adjustable parameters	config/config.toml	included
downloaded datasets	.datasets/	HuggingFace
downloaded models	/models root folder	HuggingFace
trained models	/models date-numbered subfolders	generated via training
training metadata	/results date-numbered subfolders	generated via training

---

Module	Summary	Purpose
negate	core module	Root source code folder. Creates CLI arguments and interprets commands.
→→ decompose	image processing	Random Resize Crop and Haar Wavelet transformations - arxiv:2511.14030
→→ extract	feature processing	Laplace/Sobel/Spectral analysis, VIT/VAE extraction, cross‑entropy loss - arxiv:2411.19417
→→ io	load / save / state	Hyperparameters, image datasets, console messages, model serialization and conversion.
→→ metrics	evaluation	Graphs, visualizations, model performance metadata, and a variety of heuristics for results interpretation.
→ inference	predictions	Detector functions to determine origin from trained model predictions.
→ train	XGBoost	PCA data transforms and gradient-boosted decision tree model training.

Research

The ubiqity of online services, connected presence, generative models, and the proliferate digital output that has accompanied these nascent developments have yielded a colossal and simultaneous disintegration of trust, judgement and ecological welfare, exacerbating prevailing struggles in all species of life. While the outcome of these deep-seated issues is beyond the means of a small group of academic researchers to determine, and while remediation efforts will require far more resources than attention alone, we have nevertheless taken pause to reconsider the consequences of our way of life while investigating the prospects of new avenues that may diminish harm.

@misc{darkshapes2026,
    author={darkshapes},
    title={negate},
    year={2026},
    primaryClass={cs.CV},
    howpublished={\url={https://github.com/darkshapes/negate}},
}