feat: add YOLOv8n ONNX model and ATC conversion script

- yolov8n.onnx (13MB) exported from ultralytics YOLOv8n
- scripts/atc_convert.sh for ONNX->OM conversion on Ascend 310B4
- Use FP16 precision for faster inference on edge device

scripts/atc_convert.sh

@@ -0,0 +1,42 @@
+#!/usr/bin/env bash
+# ATC 模型转换脚本
+# 在高性能电脑（CANN 环境）上执行
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+cd "$SCRIPT_DIR"
+
+MODEL=${MODEL:-yolov8n}
+INPUT_SIZE=${INPUT_SIZE:-640}
+OUTPUT_TYPE=${OUTPUT_TYPE:-FP16}
+
+echo "=========================================="
+echo "  ATC ONNX -> OM 模型转换"
+echo "=========================================="
+echo "模型: $MODEL.onnx"
+echo "输入尺寸: ${INPUT_SIZE}x${INPUT_SIZE}"
+echo "输出精度: $OUTPUT_TYPE"
+echo ""
+
+# 限制并行度，避免内存溢出
+export TBE_IMPL_MODE=AI_CORE
+export TE_PARALLEL_COMPILER=1
+
+atc \
+  --model=${MODEL}.onnx \
+  --framework=5 \
+  --output=${MODEL} \
+  --soc_version=Ascend310B4 \
+  --input_format=NCHW \
+  --input_shape="images:1,3,${INPUT_SIZE},${INPUT_SIZE}" \
+  --output_type=${OUTPUT_TYPE}
+
+echo ""
+if [ -f ${MODEL}.om ]; then
+    echo "转换成功!"
+    echo "输出: ${MODEL}.om"
+    ls -lh ${MODEL}.om
+else
+    echo "转换失败，请检查日志"
+    exit 1
+fi