ai-skills

Expert guide for converting ONNX models to CVIMODEL format for Sophgo CV181x TPU. Supports YOLO11/YOLO26 (detect, pose, seg, cls), BiSeNetv2 (semantic segmentation), and PP-LiteSeg (semantic segmentation). Includes tested conversion scripts, quantization tables (qtables), ION memory optimization (--quant_output), validation workflow, and complete documentation. Use when user needs to convert ONNX models to CVIMODEL, set up TPU-MLIR conversion pipeline, configure output names and quantization, optimize ION memory usage, validate conversion accuracy, or troubleshoot conversion issues.

Expert guide for CV181X/CV182X/CV180X (SG200X) multimedia development using CVI MPI API. Use this skill when working with: VI (video input/camera/ISP), VPSS (video processing/scaling/crop), VENC (H.264/H.265/JPEG encoding), VDEC (decoding), VB (video buffer pools), SYS binding, or any CVI_* API calls. Covers camera pipeline setup, offline VPSS processing, VB pool planning, and error diagnosis (ERR_VPSS_NOBUF, ERR_VB_NOBUF). API details in references/.

Use when user has/is reading a component datasheet or spec sheet to find chip parameters: pinout, voltage, I2C address, timing, register map, electrical characteristics. Trigger on PDF+chip questions. Also: 规格书, 数据手册, 芯片参数. All IC types.

Use when analyzing KiCad schematics (.kicad_sch) or Cadence OrCAD/Allegro schematics, identifying what a component or IC does, tracing nets, buses, signal paths, or power distribution, extracting subsystem topology, or reviewing hardware architecture for design review, BSP bring-up, or test planning. Trigger on requests about I2C/SPI/UART/USB buses, power trees, signal flow, component roles, root schematic selection in hierarchical designs, and Chinese terms such as 原理图分析, 拓扑提取, 器件角色, 信号流, 电源树.