Newer versions aren't open source, or at least have murky licencing.
replyAhh that’ll do it. A shame really, the later models seem to be fairly good just from my idle testing as an enthusiast.
replyMainly because YOLOv8 is well-supported by the Rockchip/RKNN toolchain.
replyThe goal here was an end-to-end RK3588S pipeline rather than comparing detector families: training/export, ONNX graph fixing, INT8 RKNN conversion, C++ postprocessing, and runtime inference across the 3 NPU cores. YOLOv8 has known-good export paths and Rockchip examples, so it was the most practical baseline.
Newer YOLO versions may be possible, but usually require more work around RKNN export compatibility.