老三是只猫-CSDN博客

原创 0022-opencv调用yolo-Fastest模型

1. 原作者地址主体代码import cv2 as cvimport argparseimport numpy as np# Initialize the parametersconfThreshold = 0.25 # Confidence thresholdnmsThreshold = 0.4 # Non-maximum suppression thresholdinpWidth = 320 # Width of network's input imageinpHeight =

2021-01-20 10:28:15 344 1

原创 0023-pytorch入门-读取自己的数据集并进行分类脚本

# -*- coding: utf-8 -*-# @Time : 2021/1/19 12:06# @Author : Johnsonfrom __future__ import print_function, divisionimport torchimport torch.nn as nnimport torch.optim as optimfrom torch.optim import lr_schedulerfrom torch.autograd import Varia

2021-01-19 12:08:09 278

原创 0021-python比较两张图片是否一致

def checkImgIsSameByMD5Hash(img1,img2): img_data1 = numpy.array(Image.open(img1)) img_data2 = numpy.array(Image.open(img2)) imgMD51 = hashlib.md5(img_data1).hexdigest() imgMD52 = hashlib.md5(img_data2).hexdigest() if imgMD51==imgMD52:

2021-01-19 12:02:40 414

原创 0020-pytorch-自定义网络进行分类

# -*- coding: utf-8 -*-"""Created on 2019/9/29 23:52@author: JohnsonEmail:593956670@qq.com@software: PyCharm"""import torchimport torch.nn as nnimport torch.optim as optimfrom torch.optim import lr_schedulerimport torchvision.transforms as tranf

2021-01-19 11:58:52 138

原创 0019-pytorch-celeba人脸属性分类模板代码

# -*- coding: utf-8 -*-"""Created on 2019/10/14 0:11@author: JohnsonEmail:593956670@qq.com@software: PyCharm"""###########################参考链接############################ https://blog.csdn.net/think_three/article/details/88930203import torchi

2021-01-19 11:29:12 824

原创 0018-pytorch-迁移学习范本-01

# -*- coding: utf-8 -*-# @Time : 2021/1/19 10:39# @Author : Johnsonimport osimport numpy as npimport torchimport torch.nn as nnfrom torch.optim import lr_schedulerimport torchvisionfrom torchvision import datasets, models, transformsfrom tor

2021-01-19 10:55:59 107 1

原创 0016-pytorch入门-简单分类模型脚本

# -*- encoding: utf-8 -*-"""@File : simpleNet.py@Time : 2019/9/26 15:14@Author : Johnson@Email : 593956670@qq.com"""# Import needed packagesimport torchimport torch.nn as nnfrom torchvision.datasets import CIFAR10from torchvision.transf

2021-01-19 10:35:57 113

原创 0017-fastAPI调用yoloV3模型

object_detectionimport cv2import sysimport numpy as npclass object_detector: def __init__(self, model, cfg, classes, inputSize=416): self.model = model self.cfg = cfg self.classes = classes self.inputSize = inputSi

2021-01-18 17:19:04 456

原创 0014-pytorch入门-二分类迁移学习实战

import torchimport torch.nn as nnimport torch.optim as optimfrom torch.optim import lr_schedulerfrom torch.utils.data import DataLoaderimport torchvisionfrom torchvision.transforms import transformsfrom torchvision import modelsfrom torchvision.mo.

2021-01-17 15:45:23 347

原创 0013-pytorch入门-自定义letnet-5网络

'''定义包含一些可学习的参数(或者叫权重)神经网络模型；在数据集上迭代；通过神经网络处理输入；计算损失(输出结果和正确值的差值大小)；将梯度反向传播回网络的参数；更新网络的参数，主要使用如下简单的更新原则：weight = weight - learning_rate * gradient'''#定义网络import torchimport torch.nn as nnimport torch.nn.functional as F'''CNN计算(H - k +2

2021-01-17 15:10:33 151

原创 0012-pytorch-入门图片分类脚本

# 数据获取pytorch提供了torchvision包封装了常见数据集的数据加载函数，除此之外，它还提供了torchvision.datasets和torch.utils.data.Dataloader实现图像数据转换功能#加载并处理数据import torchimport torchvisionimport torchvision.transforms as transformstransform = transforms.Compose( [transforms.ToTenso

2021-01-17 14:41:53 133

原创 0011-pytorch入门-数据并行

'''主要语句：model.gpu()tensor.gpu()#多GPU的情况下nn.DataParallel(model)'''import torchimport torch.nn as nnfrom torch.autograd import Variablefrom torch.utils.data import DataLoader,Dataset#一些参数input_size = 5output_size = 2batch_size = 30data_size

2021-01-16 23:49:22 91

原创 0010-flask调用keras模型

USAGE# Start the server:# python run_keras_server.py# Submit a request via cURL:# curl -X POST -F image=@dog.jpg 'http://localhost:5000/predict'# Submita a request via Python:# python simple_request.py# import the necessary packagesfro.

2021-01-16 22:45:05 184

原创 0009-flask调用pytorch模型

# -*- encoding: utf-8 -*-"""@File : flask_torch.py@Time : 2020/07/12 11:59@Author : Johnson@Email : 593956670@qq.com"""import ioimport jsonimport flaskimport torchimport torch.nn.functional as Ffrom PIL import Imagefrom torch import n

2021-01-16 22:13:57 519 2

原创 0008-目标检测-基于keras-yolov3训练自己的数据集

数据准备首先按照voc格式提前准备好文件夹，示例如下：生成训练集/测试集，train.txt/test.txt主要代码如下：import osimport randomtest_percent = 0.1 #可根据自己需要进行调整train_percent = 0.9xmlfilepath = 'Annotations'txtsavepath = 'ImageSets\Main'total_xml = os.listdir(xmlfilepath)num = len(total_

2021-01-16 09:37:15 259

原创 0006-目标检测-制作VOC数据集-生成Main下的txt文件

import osimport random#其中trainval=train+valtrainval_percent = 0.7 # 可以自己设置train_percent = 0.8 # 可以自己设置xmlfilepath = f"/Annotations" # 这里填写自己的标注目录txtsavepath = f"/ImageSets/Main"total_xml = os.listdir(xmlfilepath)num = len(total_xml)list = ran

2021-01-16 09:22:54 854

原创 0002-opencv dnn模块使用CUDA加速

opencv调用yolov4模型可以参考0001这篇文章安装opencv-contrib-python如果要用GPU加速，一定要安装这个包。不然就会报错。pip install opencv-contrib-python CUDA加速添加以下两行代码就可以net.setPreferableBackend(cv2.dnn.DNN_BACKEND_CUDA)net.setPreferableTarget(cv2.dnn.DNN_TARGET_CUDA)案例脚本import cv2 as

2021-01-15 21:31:18 1275 1

原创 0001-opencv4.4调用darknet-yoloV4模型

import cv2 as cvimport timenet = cv.dnn_DetectionModel('hand_v4-tiny.cfg', 'hand_v4-tiny_30000.weights')net.setInputSize(416, 416)net.setInputScale(1.0 / 255)net.setInputSwapRB(True)frame = cv.imread('imgs/69.jpg')with open('hand.names', 'rt') as

2021-01-15 21:05:18 507 1

原创 0003-dlib检测嘴巴是否张开

# USAGE# python detect_drowsiness.py --shape-predictor shape_predictor_68_face_landmarks.dat# python detect_drowsiness.py --shape-predictor shape_predictor_68_face_landmarks.dat --alarm alarm.wav# import the necessary packagesfrom scipy.spatial import

2021-01-09 14:39:07 840 1

原创 python-mysql链接脚本

#coding:utf-8import pymysqlimport datetimeclass face_mysql: def __init__(self): pass #设置数据库和密码 def conn_mysql(self): # db = pymysql.connect(host='localhost', port=3306, user='root', passwd='root', db='mysql', charset="utf8".

2021-01-05 19:46:36 167

原创 keras-视频分类脚本

"""@version: python3.6@author: 'Johnson'@contact: 593956670@qq.com@file: get_img.py@time: 2019/9/18 11:01"""import cv2import mathimport matplotlib.pyplot as pltimport pandas as pdfrom keras.preprocessing import image #预处理图像库import numpy as npf

2021-01-05 15:53:29 707

原创 Darknet中cfg文件说明和理解

[net] ★ [xxx]开始的行表示网络的一层，其后的内容为该层的参数配置，[net]为特殊的层，配置整个网络# Testing ★ #号开头的行为注释行，在解析cfg的文件时会忽略该行# batch=1# subdivisions=1# Trainingbatch=64 ★ 这儿batch与机器学习中的batch有少许差别，仅表示网络积累多少个样本后进行一次BP sub

2021-01-04 16:56:09 346

原创 OCR-文本检测后的文字纠正

# -*- encoding: utf-8 -*-"""@File : 文本角度检测以及角度矫正.py@Time : 2020/12/12 10:56@Author : Johnson参考链接：https://blog.csdn.net/qq_39056987/article/details/108671180"""'''1：霍夫变换预测倾斜角度2：图像二值化用于分割文本区域3：ROI区域获取；4：得到纠正后的文本区域截图具体实现如下：首先通过霍夫变换找到图片中

2020-12-12 11:20:02 774

原创 yoloV3转化为onnx，再转化为trt模型

yolov3模型部署实战weights转onnx并推理

2020-11-14 22:14:13 1027 1

原创人脸检测-将widerface标注转换为VOC格式

参考链接A版yoloV3-训练人脸检测

2020-11-14 22:05:17 388

原创 flask-TensorFlow推理

'''Coding: # -*- coding: utf-8 -*-Author: JohnsonDate: 2020-11-14 19:55:50Description: Email: 593956670@qq.com'''import tensorflow as tf import numpy as np PATH_TO_CKPT = 'models/ssdlite_mobilenet_v2.pb'def load_model(): detection_graph = t

2020-11-14 20:09:40 213

原创 OCR文字检测框的合并

OCR文字检测框的合并

2020-10-10 09:40:09 1297

原创 docker配置jupyter-notebook

docker配置jupyter-notebook

2020-09-27 18:31:54 149 1

原创人脸矫正代码

dlib里面这个人脸关键点检测实现的是CVPR2014 One Millisecond Face Alignment with an Ensemble of Regression Trees 里面的算法 python的dlib库的安装：下载dlib源码，有个setup.py文件，安装，然后把python_examples文件夹中dlib.so文件拷贝到python的site-pac...

2020-09-27 15:46:21 574

原创 vmware Ubuntu下调试Android studio 小米手机

设置Ubuntu的方法：在命令行下，输入：lsusb输出为：Bus 001 Device 004: ID 17ef:6050 LenovoBus 001 Device 003: ID 17ef:1003 Lenovo Integrated Smart Card ReaderBus 001 Device 002: ID 8087:0024 Intel Corp. Integrated Rate Matching HubBus 001 Device 001: ID 1d6b:0002 Linux

2020-09-07 16:25:21 582

原创 0005-Pytorch入门-使用torchvision自带的faster-rcnn进行目标检测

# -*- coding: utf-8 -*-"""Created on Thu Jul 30 08:47:12 2020@author: Johnson"""import cv2 as cvimport torchvisionimport torchfrom torchvision import transformsimport numpy as npwith open("coco.names") as f: #获取类别名称 coco_names = [line.str

2020-07-30 09:53:58 1488 2

原创 0004-身份证号码校验

# -*- coding: utf-8 -*-"""Spyder EditorThis is a temporary script file."""import re#Errors=['验证通过!','身份证号码位数不对!','身份证号码出生日期超出范围或含有非法字符!','身份证号码校验错误!','身份证地区非法!']def checkIdcard(idcard): Errors=['验证通过!','身份证号码位数不对!','身份证号码出生日期超出范围或含有非法字符!','身份证

2020-07-20 14:43:21 334

原创 pytorch-分类模型模板

#%env CUDA_VISIBLE_DEVICES=2import torch, torchvisionfrom torchvision import datasets, models, transformsimport torch.nn as nnimport torch.optim as optimfrom torch.utils.data import DataLoaderimport timefrom torchsummary import summaryimport nump

2020-05-26 23:28:19 528

原创目标检测-opencv-yolov3，包含训练以及测试代码

splitTrainTest# -*- coding: utf-8 -*-"""Created on 2020/5/26 22:21@author: JohnsonEmail:593956670@qq.com"""import randomimport os import subprocessimport sysdef split_data_set(image_dir): f_val = open("test.txt","w") f_train = open("tra

2020-05-26 23:25:06 369

原创 python-linux命令处理文件

import csvimport subprocessimport osrunMode = "train"classes = ["Snowman"]with open('class-descriptions-boxable.csv', mode='r') as infile: reader = csv.reader(infile) dict_list = {rows[1]:rows[0] for rows in reader}subprocess.run(['rm', '-

2020-05-26 22:21:07 171

原创 Chrome 错误代码"STATUS_INVALID_IMAGE_HASH"

https://www.pangzhan.net/2925.html

2020-04-30 20:25:59 5522

原创 python在内存中读取base64图片-skimage.io格式

import base64import skimage.ioimport matplotlib.pyplot as pltdef base64_to_rgb(base64_str): """ 默认base64中的图像为rgb，直接转换成即可 :param base64: :return: """ if isinstance(base64_s...

2020-02-26 19:36:31 1121