alebal123bal

4 ms·

1. 1.▲
   by alebal123bal 4mo ago

   Batching is definitely the right answer for some offline / throughput-only cases, but it was not the right tradeoff here. This pipeline is processing live camera frames and displaying/streaming annotated output, so latency and fra
2. 2.▲
   by alebal123bal 4mo ago

   Mainly because YOLOv8 is well-supported by the Rockchip/RKNN toolchain. The goal here was an end-to-end RK3588S pipeline rather than comparing detector families: training/export, ONNX graph fixing, INT8 RKNN conversion, C++ postpr
3. 3.▲
   by alebal123bal 4mo ago

   I built this while trying to understand how much of the RK3588S vision pipeline could be kept off the CPU. The main trick is not the YOLO model itself, but the pipeline structure: MIPI capture through the ISP, resize/color conversion t
4. 4.▲
   Show HN: Dual YOLOv8n UAV Detection on RK3588S at 42 FPS Using NPU(github.com)

   72 points by alebal123bal 4mo ago|9 comments