From 5db36db822550dd008dbc10c9af31876ce441a70 Mon Sep 17 00:00:00 2001 From: Lakshantha Dissanayake Date: Mon, 3 Feb 2025 06:17:42 -0800 Subject: [PATCH] Add note on failing RKNN inference on select Rockchip hardware (#18964) Signed-off-by: Glenn Jocher Co-authored-by: Glenn Jocher --- docs/en/integrations/rockchip-rknn.md | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/docs/en/integrations/rockchip-rknn.md b/docs/en/integrations/rockchip-rknn.md index ea5c4df7..087f942e 100644 --- a/docs/en/integrations/rockchip-rknn.md +++ b/docs/en/integrations/rockchip-rknn.md @@ -133,6 +133,12 @@ To install the required packages, run: yolo predict model='./yolo11n_rknn_model' source='https://ultralytics.com/images/bus.jpg' ``` +!!! note + + If you encounter a log message indicating that the RKNN runtime version does not match the RKNN Toolkit version and the inference fails, please replace `/usr/lib/librknnrt.so` with official [librknnrt.so file](https://github.com/airockchip/rknn-toolkit2/blob/master/rknpu2/runtime/Linux/librknn_api/aarch64/librknnrt.so). + + ![RKNN export screenshot](https://github.com/ultralytics/assets/releases/download/v0.0.0/rknn-npu-log.avif) + ## Benchmarks YOLO11 benchmarks below were run by the Ultralytics team on Radxa Rock 5B based on Rockchip RK3588 with `rknn` model format measuring speed and accuracy.