brainwurm/YOLOv10-Quantization

YOLOv10 — Quantized Object Detector

A minimal web UI for an INT8 dynamically-quantized YOLOv10n model, running end-to-end inference with onnxruntime (no torch/ultralytics dependency).

Drop in an image, adjust the confidence threshold, and see detections drawn live on a canvas overlay.

• —Model: yolov10n_int8_dynamic.onnx (INT8 dynamic quantization, ~2.7MB)
• —Inference: ONNX Runtime, CPU
• —Backend: Flask