AI Generated Image Detector

AI Generated Image Detector

Judges whether a picture is real or AI-generated, using a classifier trained on texture patches from about 50,000 real and 50,000 generated images. For anyone who wants a quick second opinion on an image.

Made by ymcmy on Hugging Face. These notes are Mindket's, written from its Space page and README; the app's own files and license are on Hugging Face.

About

What it is: This Space is an image classifier that judges whether a picture is real or AI-generated. It cuts the image into patches, sorts them by how much texture they hold, then runs a frequency measure on each patch before a trained classifier decides. The trained weights are included in the repository.

Who it suits: Journalists, researchers and curious users who want a second opinion on a picture, and students who want to study a small detector end to end.

Industry | Who uses it | How they use it | Example
Media | Journalists | Get a quick signal about a suspicious image | Check a picture sent in by a reader
Research | Researchers | Study a patch and frequency based detector | Read train.py and preprocess.py
Education | Students | See how an image classifier is trained | Follow the preprocess then train steps

People who need proof for a legal or newsroom decision can skip it, since a classifier gives an estimate and not proof.

Where to use it:

• A browser: the Space runs on Hugging Face with nothing to install
• Windows, macOS or Linux: run a copy of the Space with Python

Use case | Who | What they get back
Check an image | Journalists | A real or AI-generated judgment for the picture
Study the method | Researchers | Code for preprocessing, training and the model
Retrain | Students | A path to train your own copy on your data

How to set it up:

• Python with pip on Windows, macOS or Linux
• The Python packages listed in requirements.txt
• Nothing at all if you only use the Space in a browser

1. Install the Hugging Face hub package.
2. Download the Space's files into a folder.
3. Move into that folder.
4. Install Gradio 4.36.1 and the packages from requirements.txt.
5. Start the app.

pip install -U huggingface_hub
hf download ymcmy/AI_Generated_Image_Detector --repo-type space --local-dir AI_Generated_Image_Detector
cd AI_Generated_Image_Detector
pip install gradio==4.36.1
pip install -r requirements.txt
python app.py

How to use it: Open the Space page in a browser, or start your own copy with the command below. Then give the page a picture to check.

python app.py

Example: You add a picture you are unsure about. The app runs its classifier on the image and returns whether it looks real or AI-generated.

For newcomers and experts: If you are new, open the Space and try a picture you know is a photo, then one you know was generated. Treat the answer as a hint and not a verdict.

For experts, the detector splits each image into patches, labels them rich or poor in texture, and computes an azimuthal integral for each patch with a positional embedding. The author suggests running preprocess.py first and saving the data as .h5 files, then running train.py. Preprocessing is slow and keeps the CPU busy, so do not run it during training.

You add an image
It is cut into patches
Each patch gets a frequency measure
The classifier gives its judgment

Good to know:

• License: mit
• Cost to run: none named
• Data it reaches: the image you give it, and the trained weights stored in the Space
• Not for: proof that an image is fake, since the author reports 97.5 percent validation accuracy, so some images will be misjudged

On Hugging Face, 13 November 2026

• Likes: 11
• Built with: gradio
• License: mit

Links

• Space: https://huggingface.co/spaces/ymcmy/AI_Generated_Image_Detector
• Files: https://huggingface.co/spaces/ymcmy/AI_Generated_Image_Detector/tree/main

Try it

Open the Space in your browser, or copy it to your own Hugging Face account from its page. Nothing in this kit runs by itself.