AI Model Hub

All AI models used by OptiPix run entirely in your browser. No server-side processing.

How It Works

OptiPix uses Hugging Face Transformers.js and ONNX Runtime Web to run machine learning models directly in your browser using WebAssembly. Models are downloaded once and cached for offline use. Your images are never sent to any server.

SegFormer B0

Xenova/segformer-b0-finetuned-ade-512-512

Background Remover

Semantic segmentation model for separating foreground subjects from backgrounds.

Task: Image SegmentationSize: ~15 MBLicense: Apache 2.0

View on Hugging Face

ViT Base Patch16

Xenova/vit-base-patch16-224

Image Classifier

Vision Transformer model for classifying images into 1000 categories from ImageNet.

Task: Image ClassificationSize: ~86 MBLicense: Apache 2.0

View on Hugging Face

ViT-GPT2 Image Captioning

Xenova/vit-gpt2-image-captioning

Image Captioner

Vision encoder + language decoder model that generates natural language descriptions of images.

Task: Image to TextSize: ~240 MBLicense: Apache 2.0

View on Hugging Face

Tesseract.js

tesseract.js

OCR Text Extractor

WebAssembly port of the Tesseract OCR engine supporting 100+ languages.

Task: OCR (Optical Character Recognition)Size: ~2-15 MB per languageLicense: Apache 2.0