YOLOX-Tiny-ONNX

Renesas/YOLOX-Tiny-ONNX

YOLOX-Tiny-ONNX is a object detection model from Renesas. Use it when you need objects located in an image. The card lists the license as apache-2.0.

This repository hosts YOLOX-Tiny, targeting the Renesas R-Car X5H platform for object-detection inference on the NPX6 NPU.

Downloads · 30 days

Access

Public

Updated Sep 28, 2026

Repo size

20.3 MB

Likes

Public

Repo makeup

Click a slice to open those files.

.onnx20.3 MB · 100%

From the Hugging Face model README

This repository hosts YOLOX-Tiny, targeting the Renesas R-Car X5H platform for object-detection inference on the NPX6 NPU.

Model Architecture: YOLOX (Tiny variant) — an anchor-free evolution of the YOLO detector family using decoupled classification/regression heads and SimOTA label assignment, at the compact "Tiny" scale
Source Model: open-mmlab/mmyolo
Task: object-detection (dataset: coco)

The FP32 ONNX model is auto-cast to INT8 by the Renesas MWMX toolchain at compile time — no separate quantization step is required.

yolox_tiny_fast_8xb8-300e_coco.onnx (FP32)
        │
        └─▶  MWMX Runtime  ──▶  INT8 auto-cast  ──▶  NPX6 NPU

Artifact	Status	Notes
FP32 (ONNX)	⏳ Pending	`fp32/yolox_tiny_fast_8xb8-300e_coco.onnx` — to be added; will be auto-cast to INT8 by the MWMX toolchain at compile time (see Deployment Flow above); no separate INT8 file will be shipped

Measured on Renesas R-Car X5H via the MWMX runtime (APM80 ship-performance CI pipeline).

Benchmark configuration: Single NPU · Batch size: 1 · Input: 3 × 416 × 416

AI Cores	Runtime	Precision	Device	Latency (ms)	Type
1	MWMX Runtime	INT8 (auto)	X5H · 1× NPU · 1 Core · 850 MHz	3.57	Measured
3	MWMX Runtime	INT8 (auto)	X5H · 1× NPU · 3 Core · 850 MHz	2.21	Measured
4	MWMX Runtime	INT8 (auto)	X5H · 1× NPU · 4 Core · 850 MHz	2.15	Measured
6	MWMX Runtime	INT8 (auto)	X5H · 1× NPU · 6 Core · 850 MHz	2.09	Measured
12	MWMX Runtime	INT8 (auto)	X5H · 1× NPU · 12 Core · 850 MHz	2.59	Measured

TBD — not yet measured/published for this repo.

To run inference on Renesas R-Car X5H, you need:

hf download Renesas/YOLOX-Tiny-ONNX --repo-type=model --include "fp32/*"

HIL runs: Hardware-in-the-loop — measured on physical R-Car X5H silicon via the MWMX runtime (metawaremx_runtime CI pipeline, "APM80" ship-performance target)
Precision: FP32 ONNX input; INT8 execution (auto-cast by MWMX)