web-txt2img
Generate images from text prompts directly in the browser using open-weights AI models
@siva-sub/client-ocr-app
Multi-engine browser OCR with visual results. Choose between fast mobile models (PPU) or accurate server models (OnnxOCR). Features Tesseract fallback, visual bounding boxes, 100% client-side processing, PWA support, and intelligent model caching.
peekr
Browser-based ONNX eye tracking module using MediaPipe + onnxruntime-web
minisam
Plug and play ONNX SAM segmentation in the browser