1.添加模型转换脚本\n2.添加moniter.sh监控系统资源脚本

moniter.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+while true; do
+    clear
+    echo "CPU 使用率: $(top -bn1 | grep "Cpu(s)" | sed "s/.*, *\([0-9.]*\)%* id.*/\1/" | awk '{print 100 - $1}')%"
+    echo "内存使用率: $(free -m | awk 'NR==2{printf "%.2f%%\n", $3 * 100/$2 }')"
+    sleep 1
+done

transfer/001pt转onnx.py

@@ -0,0 +1,16 @@
+from ultralytics import YOLO
+import torch
+# 加载模型时禁用数据验证
+model = YOLO("/home/admin-root/haotian/康达瑞贝斯机器狗/YoloV8Obj/dataset_20250819/train2/weights/best.pt", task="detect")
+# 手动设置模型为推理模式
+model.model.eval()
+# 导出 ONNX
+dummy_input = torch.randn(1, 3, 640, 640)
+torch.onnx.export(
+    model.model,  # 使用 model.model 访问底层 PyTorch 模型
+    dummy_input,
+    "yolov8_20250820.onnx",
+    input_names=["input"],
+    output_names=["output"],
+    opset_version=11,
+)

transfer/002onnx转rknn.py

@@ -0,0 +1,58 @@
+from rknn.api import RKNN
+import cv2
+import numpy as np
+
+
+# 初始化 RKNN
+rknn = RKNN()
+
+# 配置参数（关键！）
+rknn.config(
+    target_platform="rk3588",          # 根据实际芯片型号修改
+    mean_values=[[0, 0, 0]],       # YOLOv8 输入为 0-255，无需归一化
+    std_values=[[255, 255, 255]],  # 输入数据除以 255（即 0-1 范围）
+    # quant_img_RGB2BGR=True,
+    optimization_level=3,          # 最高优化级别
+)
+
+# 加载 ONNX
+ret = rknn.load_onnx(model="/home/orangepi/Desktop/kangda_robotic_dog/yolov8_20250820.onnx")
+assert ret == 0, "加载 ONNX 失败！"
+
+# 转换模型
+ret = rknn.build(
+    do_quantization=False,          # 启用量化
+    # dataset="dataset.txt",         # 校准数据路径
+)
+assert ret == 0, "转换 RKNN 失败！"
+
+# 导出 RKNN
+ret = rknn.export_rknn("/home/orangepi/Desktop/kangda_robotic_dog/yolov8_20250820.rknn")
+assert ret == 0, "导出 RKNN 失败！"
+
+
+# Set inputs
+img = cv2.imread('/home/orangepi/Desktop/kangda_robotic_dog/微信图片_20250827165826.jpg')
+img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+img.resize((3, 640, 640))
+img = np.expand_dims(img, 0)
+
+# Init runtime environment
+print('--> Init runtime environment')
+ret = rknn.init_runtime()
+if ret != 0:
+    print('Init runtime environment failed!')
+    exit(ret)
+print('done')
+
+# Inference
+print('--> Running model')
+outputs = rknn.inference(inputs=[img], data_format=['nchw'])
+np.save('./tflite_mobilenet_v1_0.npy', outputs[0])
+print(len(outputs))
+print('done')
+
+rknn.release()
+
+
+