Arduino UNO Q Implements Model Cascading OCR

An author ports an Edge Impulse two-stage OCR pipeline to the Arduino UNO Q (Qualcomm Dragonwing QRB2210), deploying PaddleOCR detector and recognizer ONNX models via Edge Impulse BYOM. The workflow runs a lightweight detector at high FPS and conditionally invokes a heavier recognizer on cropped bounding boxes to reduce FLOPs. The proof-of-concept is runnable via a Python web UI but remains slow on the UNO Q.