Gender-Classifier-Mini

Gender-Classifier-Mini

Sorts a photo into two classes, female or male, with a fine-tuned SigLIP 2 image model. Made for developers and researchers who want a small Python classifier for photo sets.

Made by prithivMLmods on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: Gender-Classifier-Mini looks at a photo and says whether it fits a female or a male class. It is a SigLIP 2 vision-language encoder adapted for two-class image classification, and you run it through the transformers library. On a 5,000 image test set, the card gives an accuracy of 0.9720.

Who it suits: It suits developers and researchers who want a small, ready two-class image labeler in Python, and teams that need to count how a photo set splits between the two classes.

Industry | Who uses it | How they use it | Example
Research | Data analysts | Check how female and male images are spread across a dataset | Counting each class in a folder of portraits
Marketing | Campaign teams | Group audience photos by predicted class | Splitting a photo library into two sets
Software | Developers | Add a two-class image label to an app | A small Gradio page that scores an uploaded photo
Healthcare | Imaging researchers | Run gender-based analysis on medical images | Sorting scans into two groups before a study

You can skip it when you need more than two labels, or when the result affects a person's life, because a photo score is only an estimate.

Where to use it:

• Python scripts and notebooks with the transformers library
• A local Gradio page, as the card's demo script shows
• Windows, macOS or Linux with Python installed

Use case | Who | What they get back
Dataset balance check | Researchers | A count of female and male images
Quick demo page | Developers | A small page that returns two scores for an uploaded photo
Audience grouping | Marketers | Photos tagged with a predicted class

How to set it up:

• Python on Windows, macOS or Linux
• The huggingface_hub package, to download the model files
• The transformers, torch, pillow and gradio packages, to run the card's demo

1. Install the Hugging Face hub package.
2. Download the model files to your computer.
3. Install the libraries the card lists for running the model.

pip install -U huggingface_hub
hf download prithivMLmods/Gender-Classifier-Mini

!pip install -q transformers torch pillow gradio

How to use it: Day to day you load the model and its image processor by name, pass a photo through them, and read two scores, one for each class. These lines from the card do the loading.

from transformers import AutoImageProcessor
from transformers import SiglipForImageClassification
model_name = "prithivMLmods/Gender-Classifier-Mini"
model = SiglipForImageClassification.from_pretrained(model_name)
processor = AutoImageProcessor.from_pretrained(model_name)

The card also gives a longer script that wraps the model in a Gradio upload page.

Tool | What it does
SiglipForImageClassification | Loads the model with its two-class output
AutoImageProcessor | Prepares each photo for the model
gradio | Wraps the model in a page where you upload a photo

Example: You open the Gradio page from the card and upload one portrait. The page returns a score for Female ♀ and a score for Male ♂, each rounded to three decimals, and the two add up to about one.

For newcomers and experts: If you are new to this, start small. Download the files with the two commands above, run the card's demo script, and upload a single photo to see the two scores.

If you know the field, the model is a SigLIP 2 base encoder with a classification head, 92.9M parameters, about 372 MB in F32 and stored as Safetensors. It turns two logits into scores with a softmax. The trade-off is a fixed two-class label set, and the card does not describe the training data.

A photo goes in
The processor prepares the image
The encoder reads it
The head gives two logits
A softmax turns them into scores

Good to know:

Title: Test scores for each class
Female precision: 0.9660 score
Male precision: 0.9785 score
Female recall: 0.9796 score
Male recall: 0.9641 score

• Size: 92.9M parameters, about 372 MB in F32, or about 46 MB at 4-bit as an estimate
• Two classes only: Female ♀ and Male ♂
• License: apache-2.0
• Cost to run: none named, it runs on your own computer
• Data it reaches: the photos you pass in, and the model files you download from Hugging Face
• Not for: labels beyond the two classes, since the card names no others

On Hugging Face, 13 November 2026

• Task: image classification
• Library: transformers
• Parameters: 92.9M
• Weights: about 372 MB in F32; about 46 MB at 4-bit (estimate)
• Formats: Safetensors
• Downloads in the last 30 days: 6,810,513
• Likes: 20
• License: apache-2.0
• Languages: en
• Based on: google/siglip2-base-patch16-224
• Quantized versions: 2
• Tags: transformers, safetensors, siglip, image-classification, gender, male, female, siglip2

Links

• Model: https://huggingface.co/prithivMLmods/Gender-Classifier-Mini
• Files: https://huggingface.co/prithivMLmods/Gender-Classifier-Mini/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download prithivMLmods/Gender-Classifier-Mini

Then follow the usage in its model card. Nothing in this kit runs by itself.