目录
实物图:
高度集成CPU、AI计算、ISP、图形输出等功能,具有20TOPS AI算力,可以有效实现目标识别、图像分类等AI应用加速,可快速提升开发效率,降低开发成本。
可配24GB运行内存,各种复杂应用都能流畅运行。
概述
OrangePi AIpro(20T)采用昇腾AI技术路线,具体为4核64位处理器+AI处理器,集成图形处理器,支持20TOPS AI算力,拥有12GB/24GB LPDDR4X,可以外接32GB/64GB/256GB eMMC模块,支持双4K高清输出。
OrangePi AIpro(20T)引用了相当丰富的接口,包括两个HDMI输出、GPIO接口、Type-C电源接口、支持SATA/NVMe SSD 2280的M.2插槽、TF插槽、2.5G高速网口、三个USB3.0、一个USB Type-C 3.0、一个Type-C(串口打印调试功能)、两个MIPI摄像头、一个MIPI屏等,预留电池接口,可广泛适用于AI边缘计算、深度视觉学习及视频流AI分析、视频图像分析、自然语言处理、智能小车、机械臂、人工智能、无人机、云计算、AR/VR、智能安防、智能家居等领域,覆盖 AIoT各个行业。
OrangePi AIpro支持Ubuntu、openEuler操作系统,满足大多数AI算法原型验证、推理应用开发的需求。
硬件
发板搭载的是一颗64位4核CPU,具体型号不详
昇腾芯片 NPU 卡的信息,Device为310B4,芯片温度为50度
开发板使用32G内存卡作为硬盘
8G内存
时间有点紧,整理后会将源码发布
项目介绍
本文主要是介绍在OrangePi AIpro 香橙派中如何使用YOLOv5,只介绍如何使用,数据收集和标注,模型训练,模型优化在之前文章已经做过了就不再介绍.由于昨天弄的比较匆忙,忘记了把如何配置网络等步骤记录下来,下面就给大家展示一下简单的步骤,后期有时间再去格式化重新截图.香橙派的效果是非常令我震惊的.完全超乎我的想象...
基于YOLO系列算法的水果识别系统在实时目标检测领域展现了卓越的性能。YOLOv5、YOLOv6、YOLOv7和YOLOv8作为该系列的最新版本,通过不断优化网络结构和训练策略,显著提升了目标检测的精度和效率。本文旨在研究和实现一个基于YOLO系列算法的水果识别系统,以提高水果识别的准确性和实时性。本文详述了国内外研究现状、数据集处理、算法原理、模型构建与训练代码等,基于这些先进的YOLO算法,设计并实现了一个水果识别系统,能够在各种硬件平台上高效运行。系统通过摄像头实时捕捉水果图像,并利用YOLO模型进行目标检测和分类。实验结果表明,该系统在识别精度和推理速度方面均表现出色,能够满足实际应用中的需求。
实现步骤:
- 数据收集和标注:收集大量水果图像并进行手动或自动标注。
- 模型训练:使用YOLOv5的预训练模型或自定义数据集进行模型训练,以学习水果的特征和边界框。
- 模型优化:通过调整训练策略、数据增强和超参数优化,提升模型的准确率和泛化能力。
- 部署和测试:将训练好的模型部署到目标环境中,评估其在实际场景中的性能和效果。
本文主要是介绍在OrangePi AIpro 香橙派中如何使用YOLOv5,只介绍如何使用,数据收集和标注,模型训练,模型优化在之前文章已经做过了就不再介绍
环境安装
# YOLOv5 requirements
# Usage: pip install -r requirements.txt# Base ----------------------------------------
matplotlib>=3.2.2
numpy>=1.18.5
opencv-python>=4.1.1
Pillow>=7.1.2
PyYAML>=5.3.1
requests>=2.23.0
scipy>=1.4.1
torch==1.9.1
torchvision==0.10.1
pyqt5==5.15.6
tqdm>=4.64.0
# protobuf<=3.20.1 # https://github.com/ultralytics/yolov5/issues/8012# Logging -------------------------------------
tensorboard>=2.4.1
# wandb
# clearml# Plotting ------------------------------------
pandas>=1.1.4
seaborn>=0.11.0# Export --------------------------------------
# coremltools>=5.2 # CoreML export
# onnx>=1.9.0 # ONNX export
# onnx-simplifier>=0.4.1 # ONNX simplifier
# nvidia-pyindex # TensorRT export
# nvidia-tensorrt # TensorRT export
# scikit-learn==0.19.2 # CoreML quantization
# tensorflow>=2.4.1 # TFLite export (or tensorflow-cpu, tensorflow-aarch64)
# tensorflowjs>=3.9.0 # TF.js export
# openvino-dev # OpenVINO export# Extras --------------------------------------
ipython # interactive notebook
psutil # system utilization
thop>=0.1.1 # FLOPs computation
# albumentations>=1.0.3
# pycocotools>=2.0 # COCO mAP
# roboflow
直接在终端安装requirements.txt 中所需的环境.
但是其中有的环境出现报错,需要单独安装!!!
# torch_npu由于需要源码编译,速度可能较慢,本样例提供 python3.9,torch2.1版本的torch_npu whl包 wget https://obs-9be7.obs.cn-east-2.myhuaweicloud.com/wanzutao/torch_npu-2.1.0rc1-cp39-cp39-linux_aarch64.whl # 使用pip命令安装 pip3 install torch_npu-2.1.0rc1-cp39-cp39-linux_aarch64.whl
因为第一次接触,有些东西不会操作.后续操作都在JupterLab中进行.
将程序下载后并解压.
unrar x YOLOv5水果检测.rar
在JupterLab中打开
可以直接run main.py,或者可以像我一样将代码复制到.ipynb文件里运行.
结果
运行之后结果
香橙派AIpro展示出了强大的算力,能够以0.01秒的速度完成YOLOv5水果识别项目,这确实显示了其在处理复杂计算任务上的优秀性能。这种快速的识别速度反映了其高效的处理能力和优秀的硬件性能,使其能够在实时应用和大规模数据处理中发挥重要作用。
我的台式机4060ti跑这个项目视频每一帧的速度大约是0.005s,香橙派作为一个千元的机子,已经很优秀了.我完全没有想到会有这么强.
项目源码
# -*- coding: UTF-8 -*-
import random
import sys
import threading
import time
import cv2
import numpy
import torch
import torch.backends.cudnn as cudnn
from PyQt5.QtCore import *
from PyQt5.QtGui import *
from PyQt5.QtWidgets import *
from models.experimental import attempt_load
from utils.datasets import LoadImages, LoadStreams
from utils.general import check_img_size, non_max_suppression, scale_coords
from utils.plots import plot_one_box
from utils.torch_utils import select_device, time_synchronized
model_path = 'runs/train/yolov5s/weights/best.pt'
# 添加一个关于界面
# 窗口主类
class MainWindow(QTabWidget):
# 基本配置不动,然后只动第三个界面
def __init__(self):
# 初始化界面
super().__init__()
self.setWindowTitle('Yolov5检测系统')
self.resize(1200, 800)
self.setWindowIcon(QIcon("./UI/xf.jpg"))
# 图片读取进程
self.output_size = 480
self.img2predict = ""
# 空字符串会自己进行选择,首选cuda
self.device = ''
# # 初始化视频读取线程
self.vid_source = '0' # 初始设置为摄像头
# 检测视频的线程
self.threading = None
# 是否跳出当前循环的线程
self.jump_threading: bool = False
self.image_size = 640
self.confidence = 0.25
self.iou_threshold = 0.45
# 指明模型加载的位置的设备
self.model = self.model_load(weights=model_path,
device=self.device)
self.initUI()
self.reset_vid()
@torch.no_grad()
def model_load(self,
weights="", # model.pt path(s)
device='', # cuda device, i.e. 0 or 0,1,2,3 or cpu
):
"""
模型初始化
"""
device = self.device = select_device(device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, device) # load FP32 model
self.stride = int(model.stride.max()) # model stride
self.image_size = check_img_size(self.image_size, s=self.stride) # check img_size
if half:
model.half() # to FP16
# Run inference
if device.type != 'cpu':
print("Run inference")
model(torch.zeros(1, 3, self.image_size, self.image_size).to(device).type_as(
next(model.parameters()))) # run once
print("模型加载完成!")
return model
def reset_vid(self):
"""
界面重置事件
"""
self.webcam_detection_btn.setEnabled(True)
self.mp4_detection_btn.setEnabled(True)
self.left_vid_img.setPixmap(QPixmap("./UI/up.jpeg"))
self.vid_source = '0'
self.disable_btn(self.det_img_button)
self.disable_btn(self.vid_start_stop_btn)
self.jump_threading = False
def initUI(self):
"""
界面初始化
"""
# 图片检测子界面
font_title = QFont('楷体', 16)
font_main = QFont('楷体', 14)
font_general = QFont('楷体', 10)
# 图片识别界面, 两个按钮,上传图片和显示结果
img_detection_widget = QWidget()
img_detection_layout = QVBoxLayout()
img_detection_title = QLabel("图片识别功能")
img_detection_title.setFont(font_title)
mid_img_widget = QWidget()
mid_img_layout = QHBoxLayout()
self.left_img = QLabel()
self.right_img = QLabel()
self.left_img.setPixmap(QPixmap("./UI/up.jpeg"))
self.right_img.setPixmap(QPixmap("./UI/right.jpeg"))
self.left_img.setAlignment(Qt.AlignCenter)
self.right_img.setAlignment(Qt.AlignCenter)
self.left_img.setMinimumSize(480, 480)
self.left_img.setStyleSheet("QLabel{background-color: #f6f8fa;}")
mid_img_layout.addWidget(self.left_img)
self.right_img.setMinimumSize(480, 480)
self.right_img.setStyleSheet("QLabel{background-color: #f6f8fa;}")
mid_img_layout.addStretch(0)
mid_img_layout.addWidget(self.right_img)
mid_img_widget.setLayout(mid_img_layout)
self.up_img_button = QPushButton("上传图片")
self.det_img_button = QPushButton("开始检测")
self.up_img_button.clicked.connect(self.upload_img)
self.det_img_button.clicked.connect(self.detect_img)
self.up_img_button.setFont(font_main)
self.det_img_button.setFont(font_main)
self.up_img_button.setStyleSheet("QPushButton{color:white}"
"QPushButton:hover{background-color: rgb(2,110,180);}"
"QPushButton{background-color:rgb(48,124,208)}"
"QPushButton{border:2px}"
"QPushButton{border-radius:5px}"
"QPushButton{padding:5px 5px}"
"QPushButton{margin:5px 5px}")
self.det_img_button.setStyleSheet("QPushButton{color:white}"
"QPushButton:hover{background-color: rgb(2,110,180);}"
"QPushButton{background-color:rgb(48,124,208)}"
"QPushButton{border:2px}"
"QPushButton{border-radius:5px}"
"QPushButton{padding:5px 5px}"
"QPushButton{margin:5px 5px}")
img_detection_layout.addWidget(img_detection_title, alignment=Qt.AlignCenter)
img_detection_layout.addWidget(mid_img_widget, alignment=Qt.AlignCenter)
img_detection_layout.addWidget(self.up_img_button)
img_detection_layout.addWidget(self.det_img_button)
img_detection_widget.setLayout(img_detection_layout)
# 视频识别界面
# 视频识别界面的逻辑比较简单,基本就从上到下的逻辑
vid_detection_widget = QWidget()
vid_detection_layout = QVBoxLayout()
vid_title = QLabel("视频检测功能")
vid_title.setFont(font_title)
self.left_vid_img = QLabel()
self.right_vid_img = QLabel()
self.left_vid_img.setPixmap(QPixmap("./UI/up.jpeg"))
self.right_vid_img.setPixmap(QPixmap("./UI/right.jpeg"))
self.left_vid_img.setAlignment(Qt.AlignCenter)
self.left_vid_img.setMinimumSize(480, 480)
self.left_vid_img.setStyleSheet("QLabel{background-color: #f6f8fa;}")
self.right_vid_img.setAlignment(Qt.AlignCenter)
self.right_vid_img.setMinimumSize(480, 480)
self.right_vid_img.setStyleSheet("QLabel{background-color: #f6f8fa;}")
mid_img_widget = QWidget()
mid_img_layout = QHBoxLayout()
mid_img_layout.addWidget(self.left_vid_img)
mid_img_layout.addStretch(0)
mid_img_layout.addWidget(self.right_vid_img)
mid_img_widget.setLayout(mid_img_layout)
self.webcam_detection_btn = QPushButton("摄像头实时监测")
self.mp4_detection_btn = QPushButton("视频文件检测")
self.vid_start_stop_btn = QPushButton("启动/停止检测")
self.webcam_detection_btn.setFont(font_main)
self.mp4_detection_btn.setFont(font_main)
self.vid_start_stop_btn.setFont(font_main)
self.webcam_detection_btn.setStyleSheet("QPushButton{color:white}"
"QPushButton:hover{background-color: rgb(2,110,180);}"
"QPushButton{background-color:rgb(48,124,208)}"
"QPushButton{border:2px}"
"QPushButton{border-radius:5px}"
"QPushButton{padding:5px 5px}"
"QPushButton{margin:5px 5px}")
self.mp4_detection_btn.setStyleSheet("QPushButton{color:white}"
"QPushButton:hover{background-color: rgb(2,110,180);}"
"QPushButton{background-color:rgb(48,124,208)}"
"QPushButton{border:1px}"
"QPushButton{border-radius:5px}"
"QPushButton{padding:5px 5px}"
"QPushButton{margin:5px 5px}")
self.vid_start_stop_btn.setStyleSheet("QPushButton{color:white}"
"QPushButton:hover{background-color: rgb(2,110,180);}"
"QPushButton{background-color:rgb(48,124,208)}"
"QPushButton{border:2px}"
"QPushButton{border-radius:5px}"
"QPushButton{padding:5px 5px}"
"QPushButton{margin:5px 5px}")
self.webcam_detection_btn.clicked.connect(self.open_cam)
self.mp4_detection_btn.clicked.connect(self.open_mp4)
self.vid_start_stop_btn.clicked.connect(self.start_or_stop)
# 添加fps显示
fps_container = QWidget()
fps_container.setStyleSheet("QWidget{background-color: #f6f8fa;}")
fps_container_layout = QHBoxLayout()
fps_container.setLayout(fps_container_layout)
# 左容器
fps_left_container = QWidget()
fps_left_container.setStyleSheet("QWidget{background-color: #f6f8fa;}")
fps_left_container_layout = QHBoxLayout()
fps_left_container.setLayout(fps_left_container_layout)
# 右容器
fps_right_container = QWidget()
fps_right_container.setStyleSheet("QWidget{background-color: #f6f8fa;}")
fps_right_container_layout = QHBoxLayout()
fps_right_container.setLayout(fps_right_container_layout)
# 将左容器和右容器添加到fps_container_layout中
fps_container_layout.addWidget(fps_left_container)
fps_container_layout.addStretch(0)
fps_container_layout.addWidget(fps_right_container)
# 左容器中添加fps显示
raw_fps_label = QLabel("原始帧率:")
raw_fps_label.setFont(font_general)
raw_fps_label.setAlignment(Qt.AlignLeft)
raw_fps_label.setStyleSheet("QLabel{margin-left:80px}")
self.raw_fps_value = QLabel("0")
self.raw_fps_value.setFont(font_general)
self.raw_fps_value.setAlignment(Qt.AlignLeft)
fps_left_container_layout.addWidget(raw_fps_label)
fps_left_container_layout.addWidget(self.raw_fps_value)
# 右容器中添加fps显示
detect_fps_label = QLabel("检测帧率:")
detect_fps_label.setFont(font_general)
detect_fps_label.setAlignment(Qt.AlignRight)
self.detect_fps_value = QLabel("0")
self.detect_fps_value.setFont(font_general)
self.detect_fps_value.setAlignment(Qt.AlignRight)
self.detect_fps_value.setStyleSheet("QLabel{margin-right:96px}")
fps_right_container_layout.addWidget(detect_fps_label)
fps_right_container_layout.addWidget(self.detect_fps_value)
# 添加组件到布局上
vid_detection_layout.addWidget(vid_title, alignment=Qt.AlignCenter)
vid_detection_layout.addWidget(fps_container)
vid_detection_layout.addWidget(mid_img_widget, alignment=Qt.AlignCenter)
vid_detection_layout.addWidget(self.webcam_detection_btn)
vid_detection_layout.addWidget(self.mp4_detection_btn)
vid_detection_layout.addWidget(self.vid_start_stop_btn)
vid_detection_widget.setLayout(vid_detection_layout)
# 关于界面
about_widget = QWidget()
about_layout = QVBoxLayout()
about_title = QLabel('欢迎使用目标检测系统\n') # 修改欢迎词语
about_title.setFont(QFont('楷体', 18))
about_title.setAlignment(Qt.AlignCenter)
about_img = QLabel()
about_img.setPixmap(QPixmap('./UI/qq.png'))
about_img.setAlignment(Qt.AlignCenter)
# label4.setText("<a href='https://oi.wiki/wiki/学习率的调整'>如何调整学习率</a>")
label_super = QLabel() # 更换作者信息
label_super.setText("")
label_super.setFont(QFont('楷体', 16))
label_super.setOpenExternalLinks(True)
# label_super.setOpenExternalLinks(True)
label_super.setAlignment(Qt.AlignRight)
about_layout.addWidget(about_title)
about_layout.addStretch()
about_layout.addWidget(about_img)
about_layout.addStretch()
about_layout.addWidget(label_super)
about_widget.setLayout(about_layout)
self.addTab(img_detection_widget, '图片检测')
self.addTab(vid_detection_widget, '视频检测')
# self.addTab(about_widget, '联系我')
self.setTabIcon(0, QIcon('./UI/lufei.png'))
self.setTabIcon(1, QIcon('./UI/lufei.png'))
def disable_btn(self, pushButton: QPushButton):
pushButton.setDisabled(True)
pushButton.setStyleSheet("QPushButton{background-color: rgb(2,110,180);}")
def enable_btn(self, pushButton: QPushButton):
pushButton.setEnabled(True)
pushButton.setStyleSheet(
"QPushButton{background-color: rgb(48,124,208);}"
"QPushButton{color:white}"
)
def detect(self, source: str, left_img: QLabel, right_img: QLabel):
"""
@param source: file/dir/URL/glob, 0 for webcam
@param left_img: 将左侧QLabel对象传入,用于显示图片
@param right_img: 将右侧QLabel对象传入,用于显示图片
"""
model = self.model
img_size = [self.image_size, self.image_size] # inference size (pixels)
conf_threshold = self.confidence # confidence threshold
iou_threshold = self.iou_threshold # NMS IOU threshold
device = self.device # cuda device, i.e. 0 or 0,1,2,3 or cpu
classes = None # filter by class: --class 0, or --class 0 2 3
agnostic_nms = False # class-agnostic NMS
augment = False # augmented inference
half = device.type != 'cpu' # half precision only supported on CUDA
if source == "":
self.disable_btn(self.det_img_button)
QMessageBox.warning(self, "请上传", "请先上传视频或图片再进行检测")
else:
source = str(source)
webcam = source.isnumeric()
# Set Dataloader
if webcam:
cudnn.benchmark = True # set True to speed up constant image size inference
dataset = LoadStreams(source, img_size=img_size, stride=self.stride)
else:
dataset = LoadImages(source, img_size=img_size, stride=self.stride)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
# 用来记录处理的图片数量
count = 0
# 计算帧率开始时间
fps_start_time = time.time()
for path, img, im0s, vid_cap in dataset:
# 直接跳出for,结束线程
if self.jump_threading:
# 清除状态
self.jump_threading = False
break
count += 1
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=augment)[0]
# Apply NMS
pred = non_max_suppression(pred, conf_threshold, iou_threshold, classes=classes, agnostic=agnostic_nms)
t2 = time_synchronized()
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
s, im0 = 'detect : ', im0s[i].copy()
else:
s, im0 = 'detect : ', im0s.copy()
# s += '%gx%g ' % img.shape[2:] # print string
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f"{n} {names[int(c)]}{'s' * (n > 1)}, " # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
label = f'{names[int(cls)]} {conf:.2f}'
plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=3)
if webcam or vid_cap is not None:
if webcam: # batch_size >= 1
img = im0s[i]
else:
img = im0s
img = self.resize_img(img)
img = QImage(img.data, img.shape[1], img.shape[0], img.shape[2] * img.shape[1],
QImage.Format_RGB888)
left_img.setPixmap(QPixmap.fromImage(img))
# 计算一次帧率
if count % 10 == 0:
fps = int(10 / (time.time() - fps_start_time))
self.detect_fps_value.setText(str(fps))
fps_start_time = time.time()
# 应该调整一下图片的大小
# 时间显示
timenumber = time.strftime('%Y/%m/%d/-%H:%M:%S', time.localtime(time.time()))
im0 = cv2.putText(im0, timenumber, (50, 50), cv2.FONT_HERSHEY_SIMPLEX,
1, (0, 255, 0), 2, cv2.LINE_AA)
im0 = cv2.putText(im0, s, (50, 80), cv2.FONT_HERSHEY_SIMPLEX,
1, (255, 0, 0), 2, cv2.LINE_AA)
img = self.resize_img(im0)
img = QImage(img.data, img.shape[1], img.shape[0], img.shape[2] * img.shape[1],
QImage.Format_RGB888)
right_img.setPixmap(QPixmap.fromImage(img))
# Print time (inference + NMS)
print(f'{s}Done. ({t2 - t1:.3f}s)')
# 使用完摄像头释放资源
if webcam:
for cap in dataset.caps:
cap.release()
else:
dataset.cap and dataset.cap.release()
def resize_img(self, img):
"""
调整图片大小,方便用来显示
@param img: 需要调整的图片
"""
resize_scale = min(self.output_size / img.shape[0], self.output_size / img.shape[1])
img = cv2.resize(img, (0, 0), fx=resize_scale, fy=resize_scale)
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
return img
def upload_img(self):
"""
上传图片
"""
# 选择录像文件进行读取
fileName, fileType = QFileDialog.getOpenFileName(self, 'Choose file', '', '*.jpg *.png *.tif *.jpeg')
if fileName:
self.img2predict = fileName
# 将上传照片和进行检测做成互斥的
self.enable_btn(self.det_img_button)
self.disable_btn(self.up_img_button)
# 进行左侧原图展示
img = cv2.imread(fileName)
# 应该调整一下图片的大小
img = self.resize_img(img)
img = QImage(img.data, img.shape[1], img.shape[0], img.shape[2] * img.shape[1], QImage.Format_RGB888)
self.left_img.setPixmap(QPixmap.fromImage(img))
# 上传图片之后右侧的图片重置
self.right_img.setPixmap(QPixmap("./UI/right.jpeg"))
def detect_img(self):
"""
检测图片
"""
# 重置跳出线程状态,防止其他位置使用的影响
self.jump_threading = False
self.detect(self.img2predict, self.left_img, self.right_img)
# 将上传照片和进行检测做成互斥的
self.enable_btn(self.up_img_button)
self.disable_btn(self.det_img_button)
def open_mp4(self):
"""
开启视频文件检测事件
"""
print("开启视频文件检测")
fileName, fileType = QFileDialog.getOpenFileName(self, 'Choose file', '', '*.mp4 *.avi')
if fileName:
self.disable_btn(self.webcam_detection_btn)
self.disable_btn(self.mp4_detection_btn)
self.enable_btn(self.vid_start_stop_btn)
# 生成读取视频对象
cap = cv2.VideoCapture(fileName)
# 获取视频的帧率
fps = cap.get(cv2.CAP_PROP_FPS)
# 显示原始视频帧率
self.raw_fps_value.setText(str(fps))
if cap.isOpened():
# 读取一帧用来提前左侧展示
ret, raw_img = cap.read()
cap.release()
else:
QMessageBox.warning(self, "需要重新上传", "请重新选择视频文件")
self.disable_btn(self.vid_start_stop_btn)
self.enable_btn(self.webcam_detection_btn)
self.enable_btn(self.mp4_detection_btn)
return
# 应该调整一下图片的大小
img = self.resize_img(numpy.array(raw_img))
img = QImage(img.data, img.shape[1], img.shape[0], img.shape[2] * img.shape[1], QImage.Format_RGB888)
self.left_vid_img.setPixmap(QPixmap.fromImage(img))
# 上传图片之后右侧的图片重置
self.right_vid_img.setPixmap(QPixmap("./UI/right.jpeg"))
self.vid_source = fileName
self.jump_threading = False
def open_cam(self):
"""
打开摄像头事件
"""
print("打开摄像头")
self.disable_btn(self.webcam_detection_btn)
self.disable_btn(self.mp4_detection_btn)
self.enable_btn(self.vid_start_stop_btn)
self.vid_source = "0"
self.jump_threading = False
# 生成读取视频对象
cap = cv2.VideoCapture(0)
# 获取视频的帧率
fps = cap.get(cv2.CAP_PROP_FPS)
# 显示原始视频帧率
self.raw_fps_value.setText(str(fps))
if cap.isOpened():
# 读取一帧用来提前左侧展示
ret, raw_img = cap.read()
cap.release()
else:
QMessageBox.warning(self, "需要重新上传", "请重新选择视频文件")
self.disable_btn(self.vid_start_stop_btn)
self.enable_btn(self.webcam_detection_btn)
self.enable_btn(self.mp4_detection_btn)
return
# 应该调整一下图片的大小
img = self.resize_img(numpy.array(raw_img))
img = QImage(img.data, img.shape[1], img.shape[0], img.shape[2] * img.shape[1], QImage.Format_RGB888)
self.left_vid_img.setPixmap(QPixmap.fromImage(img))
# 上传图片之后右侧的图片重置
self.right_vid_img.setPixmap(QPixmap("./UI/right.jpeg"))
def start_or_stop(self):
"""
启动或者停止事件
"""
print("启动或者停止")
if self.threading is None:
# 创造并启动一个检测视频线程
self.jump_threading = False
self.threading = threading.Thread(target=self.detect_vid)
self.threading.start()
self.disable_btn(self.webcam_detection_btn)
self.disable_btn(self.mp4_detection_btn)
else:
# 停止当前线程
# 线程属性置空,恢复状态
self.threading = None
self.jump_threading = True
self.enable_btn(self.webcam_detection_btn)
self.enable_btn(self.mp4_detection_btn)
def detect_vid(self):
"""
视频检测
视频和摄像头的主函数是一样的,不过是传入的source不同罢了
"""
print("视频开始检测")
self.detect(self.vid_source, self.left_vid_img, self.right_vid_img)
print("视频检测结束")
# 执行完进程,刷新一下和进程有关的状态,只有self.threading是None,
# 才能说明是正常结束的线程,需要被刷新状态
if self.threading is not None:
self.start_or_stop()
def closeEvent(self, event):
"""
界面关闭事件
"""
reply = QMessageBox.question(
self,
'quit',
"Are you sure?",
QMessageBox.Yes | QMessageBox.No,
QMessageBox.No
)
if reply == QMessageBox.Yes:
self.jump_threading = True
self.close()
event.accept()
else:
event.ignore()
if __name__ == "__main__":
app = QApplication(sys.argv)
mainWindow = MainWindow()
mainWindow.show()
sys.exit(app.exec_())
个人感受
OrangePi AIpro开发板是一款卓越的AI边缘计算设备,它不仅推动了国内AI部署的进程,而且其强大的算力完全能够满足商业级产品的需求。在我使用了几天后的感受是,香橙派AIpro表现出色,特别是在AI算力方面。相比于众所周知的树莓派等Linux系列开发板,它最高可达16GB的LPDDR4X内存,性能优势明显。接口丰富多样,而且官方提供的散热器和风扇静音效果极佳,风力强劲,散热效果出色。风扇支持CPU智能控制转速,这在众多开发板中独具优势。
香橙派官方提供了详尽的开发资料,特别是用户手册的详细程度令人印象深刻,极大地提升了使用体验的顺畅度。在该款开发板的核心部件中,不乏国产芯片的身影,这预示着未来国内科技和产业的进一步发展,希望能实现新的飞跃,推动国产化事业迈上新的台阶。