balitaacro/alpha-scan

Image Segmentation

Semantic segmentation with SegFormer-B0. Built with ONNX Runtime for production deployment.

Features

• —Streaming token generation
• —Model quantization support (INT8/FP16)
• —GPU and CPU inference support

API

POST /tokenize     — Tokenize input text
GET  /metrics      — Prometheus-compatible metrics
POST /predict      — Run model inference on input
GET  /health       — Liveness and readiness probe