yolov5测试单张图片,返回一个列表[类别,置信度,x,y,w,h]
from numpy import random
import torch
from models.experimental import attempt_load
from utils.datasets import LoadStreams, LoadImages
from utils.general import (
check_img_size, non_max_suppression, apply_classifier, scale_coords,
xyxy2xywh, plot_one_box, strip_optimizer, set_logging)
from utils.torch_utils import select_device, load_classifier, time_synchronized
import os
import shutil
# Initialize
out = r'inference\output'
set_logging()
device = select_device('')
if os.path.exists(out):
shutil.rmtree(out) # delete output folder
os.makedirs(out) # make new output folder
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load('weights/yolov5s.pt', map_location=device) # load FP32 model
imgsz = check_img_size(512, s=model.stride.max()) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(torch.load('weights/resnet101.pt', map_location=device)['model']) # load weights
modelc.to(device).eval()
# Set Dataloader
# 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 range(len(names))]
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img) if device.type != 'cpu' else None # run once
def PoseDect(path, imgsz=512):
res = []
dataset = LoadImages(path, img_size=imgsz)
for path, img, im0s, vid_cap in dataset:
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=False)[0]
# Apply NMS
pred = non_max_suppression(pred, 0.4, 0.5, classes=None, agnostic=False)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
gn = torch.tensor(im0s.shape)[[1, 0, 1, 0]] # normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0s.shape).round()
for *xyxy, conf, cls in reversed(det):
x, y, w, h = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist()
res.append([names[int(cls)], float(conf), x, y, w, h])
return res
# for _, img, im0s, _ in dataset:
#
# 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)
#
# pred = model(img, augment=False)[0]
# # Apply NMS
# pred = non_max_suppression(pred, 0.4, .05, classes=None, agnostic=None)
#
# for i, det in enumerate(pred): # detections per image
# p, s, im0 = path[i], '%g: ' % i, im0s[i].copy()
#
# gn = torch.tensor(im0.shape)[[1, 0, 1, 0]] # normalization gain whwh
# if det is not None and len(det):
# # Rescale boxes from img_size to im0 size
# det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
#
# # Write results
# for *xyxy, conf, cls in reversed(det):
# xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist() # normalized xywh
# x, y, w, h = xywh
# if(int(cls)==0):
# res.append([names[int(cls)], float(conf), x, y, w, h])
# #res.append([int(cls), float(conf), x, y, w, h])
#
# # draw
# # label = '%s %.2f' % (names[int(cls)], conf)
# #plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=3)
#
# return res
if __name__ == '__main__':
path = r'inference\images\0152498D-225A-4126-AEBE-B6D9423E12E7.png'
s = PoseDect(path=path)
print(s)
import cv2
img = cv2.imread(r'inference\images\0152498D-225A-4126-AEBE-B6D9423E12E7.png')
for box in s:
x1, y1, x2, y2 = box[2:]
# 映射原图尺寸
x = int(x1 * img.shape[1])
y = int(y1 * img.shape[0])
w = int(x2 * img.shape[1])
h = int(y2 * img.shape[0])
# 计算出左上角和右下角:原x,y是矩形框的中心点
a = int(x - w / 2)
b = int(y - h / 2)
c = int(x + w / 2)
d = int(y + h / 2)
print(x1, y1, x1 + x2, y1 + y2)
print(x, y, x + w, y + h)
print(a, b, c, d)
cv2.rectangle(img, (a, b), (c, d), (255, 0, 0), 2)
cv2.imshow('dst', img)
cv2.waitKey()
cv2.destroyAllWindows()
文章浏览阅读1.4k次。当前顶点状态当一个顶点数组被定义,但是数据还不可用时,当前值常用于联合顶点辅助的数据。当前值可能在任何时候被独占命令改变。使用以下的命令可设置当前的RGBA颜色值。void Color4{xf}(T red, T green, T blue, T alpha);void Color4ub(ubyte red, ubyte green, ubyte blue, uby_查看当前顶点在哪个顶点组
文章浏览阅读603次。(完结篇)实现自动行走的功能(Move类)类成员与构造函数private long time;private ImageView imageView;private int direID; double x; double maxx; double width; Rectangle2D screenBounds = Screen.getPrimary().getVisualBounds(); Stage stage;private EventListener lis
文章浏览阅读3.9w次,点赞39次,收藏182次。AnacondaAnaconda指的是一个开源的Python发行版本,其包含了conda、Python等180多个科学包及其依赖项。因为包含了大量的科学包,Anaconda的下载文件比较大(约531 MB),如果只需要某些包,或者需要节省带宽或存储空间,也可以使用Miniconda这个较小的发行版(仅包含conda和Python)。Conda是一个开源的包、环境管理器,可以用于在同一个机器上安装不同版本的软件包及其依赖,并能够在不同的环境之间切换。Anaconda包括Conda、Pyth..._anaconda环境变量
文章浏览阅读601次。代码:import pandas as pdimport numpy as npprint("生成序列")s = pd.Series([1, 3, 6, np.nan, 44, 1])print(s)dates = pd.date_range('20160101', periods=6)print(dates)运行结果:生成序列0 1.01 3.02 6..._s = pd.series(['湖北', '浙江', '广东'])
文章浏览阅读200次。最近看到手机自带的时钟App可以实时的修改图标样式,就觉得很神奇。考虑到我自己要做的日记App可能也需要这样的功能,于是研究了一下,目前初步实现了修改图标及点击后进入不同Activity的功能,代码比较简单,点击这里download ZIP。简单效果图:原来调用修改方法:代码很简单,就2个部分:1.修改manifest文件:package="sxkeji.net.aliasdemo" >an..._android app 更换电子门牌 demo
文章浏览阅读2.4w次,点赞185次,收藏333次。Pytorch中的register_buffer1.register_buffer( )的使用随着例子边看边讲例子1:使用类成员变量(类成员变量并不会在我们的model.state_dict(),即无法保存)例子2:使用类成员变量(类成员变量并不会随着model.cuda()复制到gpu上)例子3:使用register_buffer()总结2.与pa1.register_buffer( )的使用回顾模型保存:torch.save(model.state_dict()),model.state_dict(_register_buffer
文章浏览阅读1.6k次。一、创建数据库mysql> create database news character set utf8;Query OK, 0 rows affected (0.09 sec)二、创建用户mysql> create user ‘news’@‘39.15.16.14’ identified by ‘123news’;Query OK, 0 rows affected (0.09 sec)三、授权用户mysql> grant all privileges on news.* _mysql8.0 添加创建数据库权限
文章浏览阅读1.9k次。本文主要内容是sonarqube安装插件实现功能扩展,也对sonarqube的API功能进行了可用性确认。_sonar-pdf-plugin
文章浏览阅读6.4w次,点赞6次,收藏44次。基本操作传送门显示效果图添加效果图代码:<template> <div style="width:90%;margin: 20px auto;"> <!-- ref 用来拿到这个日历对象 defaultView 默认以月份方式显示 locale 语言 header 头部的按钮 buttonText 头部按钮的文字 plugins 插件 wee......_vue 行程插件
文章浏览阅读1.2w次,点赞5次,收藏8次。layout_centerHorizontal是相对于RelativeLayout的布局属性如果设置为true,就将该控价设置在相对于父控件水平居中的位置layout_gravity针对LinearLayout的一种控件对齐方式,可以把值设置成下列值:center_vertical、center_horizontal、center等等gravity控制控件内文字的对齐方式举个栗子:在写一个简单的
文章浏览阅读1.9k次。#一,循环嵌套##1,双for循环嵌套"""外层循环每循环一次,内层循环循环所有次,内层循环体执行就是内外层循环次数的乘积for i in "范围": #循环9次 for j in "范围":# 循环9次 "循环体""""for x in range(3): for y in range(3): print("%d %d" % (x, y))#例2打印99 乘法表for i in range(1, 10):# 循环打印数字0到9 for_python循环j和i每循环9个一分行
文章浏览阅读9.6k次,点赞2次,收藏6次。ubuntu笔记本 安装百度网盘 ,帮助完成安装 百度网盘 的资料来源,请点击此处hairui@hadoop:~$ git clone https://gitee.com/wszqkzqk/deepin-wine-for-ubuntu.git正克隆到 'deepin-wine-for-ubuntu'...remote: Enumerating objects: 2777, done.remo..._ubuntu kylin16.04 网盘