之前通过百度AI接口实现了图像识别，目标做图像识别再进行垃圾分类提示，于是乎我在网上查询各垃圾分类的数据集，很多数据集收费的各大网站让我很反感，接下来放两个比较nice的开源站：（开源让技术进步！）
百度AI社区 | 上传者：Thomas-yanxin，快速链接： 垃圾分类数据集ImageNet格式，用于训练效果是很不错的
知乎 | 作者：谢伟通过go语言实现了后端查询垃圾分类，其中包含垃圾分类数据集csv格式。
ps.后文使用到的数据集为csv格式，供学习使用。
正文
1.数据集
数据集包含2534条数据，大致分为4类，根据go语言4类分类，重新处理了一下数据集：
参考如下：
2.表格匹配规则
读取表格B列存储的名称和识别的名称对比，对比规则： 字符串相等 sort：1 不相等，但包含字符： 字符串相等的字节占总字节的比例 sort = len(name)/len(cell.value)
import openpyxl
from openpyxl import Workbook
def excel(excel_file, name):
# 定义列表result存储所有读取数据
result = {"id":0,"name":"test","imageUrl":"","sortId":"可回收垃圾","result":"","sort":0}
wb = openpyxl.load_workbook(excel_file) # 读取excel文件
sheet = wb.worksheets[0]#读取第一个表
col = sheet["B"]#读取B列
for index,cell in enumerate(col):
if cell.value == name:
result["sort"] = 1
result["name"] = name
result["result"] = cell.value
result["id"] = sheet.cell(index,1).value
result["sortId"] = sheet.cell(index,4).value
print(cell.value,1)
else:
if name in cell.value:
sort = len(name)/len(cell.value)
print(cell.value,sort)
if sort > result["sort"]:
result["sort"] = sort
result["name"] = name
result["result"] = cell.value
result["id"] = sheet.cell(index,1).value
result["sortId"] = sheet.cell(index,4).value
print(cell.value,sort)
if result["sort"] == 0:
print("未识别到是什么垃圾")
else:
print("====识别结果=======")
print(result)
print(result["name"],"的识别结果是：",result["result"],"是",result["sortId"],"可信度为：",result["sort"])
excel(r'/Users/wangyu/Desktop/waste.xlsx',
r'花生') # 调用函数，传入参数
3.图像识别+垃圾分类
参考上一篇文章图像识别
# coding=utf-8
import requests
import json
import base64
# 防止https证书校验不正确
import ssl
ssl._create_default_https_context = ssl._create_unverified_context
"""
获取token
"""
def getToken(AccessKey,SecretKey):
# client_id 为官网获取的AK， client_secret 为官网获取的SK
host = 'https://aip.baidubce.com/oauth/2.0/token?grant_type=client_credentials&client_id='+AccessKey+'&client_secret='+SecretKey
headers = {
'Content-Type': 'application/json;charset=UTF-8',
'grant_type': 'client_credentials',
}
access_token = ''
response = requests.get(url=host, headers=headers)
if response:
res = response.json()
access_token = res['access_token']
return access_token
def getResult(url,access_token):
# 植物
request_url = "https://aip.baidubce.com/rest/2.0/image-classify/v1/plant"
# 动物
# request_url = "https://aip.baidubce.com/rest/2.0/image-classify/v1/animal"
# 二进制方式打开图片文件
f = open(url, 'rb')
img = base64.b64encode(f.read())
params = {"image":img,"top_num":1,"baike_num":1}
request_url = request_url + "?access_token=" + access_token
headers = {'content-type': 'application/json'}
response = requests.post(request_url, data=params, headers=headers)
result = response.json()
if result:
print (json.dumps(result,indent=1,ensure_ascii=False))
# 打印图片结果
for data in result["result"]:
print(u" 菜品名称: " + data["name"])
return data["name"]
else:
return "未识别该图片"
def getWaste(name):
waste = '未识别到该垃圾的分类'
# 定义列表result存储所有读取数据
result = {"id":0,"name":"test","imageUrl":"","sortId":"可回收垃圾","result":"","sort":0}
excel_file = "/Users/wangyu/Desktop/waste.xlsx"
wb = openpyxl.load_workbook(excel_file) # 读取excel文件
sheet = wb.worksheets[0]#读取第一个表
col = sheet["B"]#读取B列
for index,cell in enumerate(col):
if cell.value == name:
result["sort"] = 1
result["name"] = name
result["result"] = cell.value
result["id"] = sheet.cell(index,1).value
result["sortId"] = sheet.cell(index,4).value
print(cell.value,1)
else:
if name in cell.value:
sort = len(name)/len(cell.value)
print(cell.value,sort)
if sort > result["sort"]:
result["sort"] = sort
result["name"] = name
result["result"] = cell.value
result["id"] = sheet.cell(index,1).value
result["sortId"] = sheet.cell(index,4).value
print(cell.value,sort)
if result["sort"] == 0:
print("未识别到是什么垃圾")
return waste
else:
print("====识别结果=======")
print(result["name"],"的识别结果是：",result["result"],"是",result["sortId"],"可信度为：",result["sort"])
return result
if __name__ == '__main__':
# 识别的图片
# url = '/Users/wangyu/Desktop/shicai.jpg'
# url = '/Users/wangyu/Desktop/fish.jpg'
url = '/Users/wangyu/Desktop/hongzao.jpg'
# 百度账号信息
AccessKey = ''
SecretKey = ''
# 获取小票识别结果
access_token = getToken(AccessKey,SecretKey)
img_result = getResult(url,access_token)
waste_result = getWaste(img_result)4.整合显示
通过标识将主要内容输出显示于图片上：
# coding=utf-8
# 网络数据请求
import requests
import json
import base64
import numpy as np
# 表格处理
import openpyxl
from openpyxl import Workbook
# 图像处理
import cv2
from matplotlib import pyplot as plt
from urllib import request
from PIL import Image, ImageDraw, ImageFont
# 防止https证书校验不正确
import ssl
ssl._create_default_https_context = ssl._create_unverified_context
"""
获取token
"""
def getToken(AccessKey,SecretKey):
# client_id 为官网获取的AK， client_secret 为官网获取的SK
host = 'https://aip.baidubce.com/oauth/2.0/token?grant_type=client_credentials&client_id='+AccessKey+'&client_secret='+SecretKey
headers = {
'Content-Type': 'application/json;charset=UTF-8',
'grant_type': 'client_credentials',
}
access_token = ''
response = requests.get(url=host, headers=headers)
if response:
res = response.json()
access_token = res['access_token']
return access_token
"""
图像识别结果
输入：本地图片地址，token
输出：识别结果，识别分数
"""
def getResult(url,access_token):
img = cv2.imread(url)
plt.figure(figsize=(5,5))
plt.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))#BGR转RGB
plt.xlabel(u'img')
plt.show()
# 植物
request_url = "https://aip.baidubce.com/rest/2.0/image-classify/v1/plant"
# 动物
# request_url = "https://aip.baidubce.com/rest/2.0/image-classify/v1/animal"
# 二进制方式打开图片文件
f = open(url, 'rb')
img = base64.b64encode(f.read())
params = {"image":img,"top_num":1,"baike_num":1}
request_url = request_url + "?access_token=" + access_token
headers = {'content-type': 'application/json'}
response = requests.post(request_url, data=params, headers=headers)
result = response.json()
if result:
print (json.dumps(result,indent=1,ensure_ascii=False))
# 打印图片结果
for data in result["result"]:
print(u" 菜品名称: " + data["name"])
if data["baike_info"]["image_url"]:
print(u" 图片为" + data["baike_info"]["image_url"])
plt.figure(figsize=(5,5))
response = requests.get(data["baike_info"]["image_url"])
resp = request.urlopen(data["baike_info"]["image_url"])
image = np.asarray(bytearray(resp.read()), dtype="uint8")
image = cv2.imdecode(image, cv2.IMREAD_COLOR)
plt.imshow(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))#BGR转RGB
plt.xlabel("score is"+str(data["score"]))
plt.show()
return data["name"],data["score"]
else:
return "未识别该图片",0
"""
垃圾分类识别
输入：物体名称
输出：垃圾分类结果，分数
"""
def getWaste(name):
waste = '未识别到该垃圾的分类'
# 定义列表result存储所有读取数据
result = {"id":0,"name":"test","imageUrl":"","sortId":"可回收垃圾","result":"","sort":0}
excel_file = "/Users/wangyu/Desktop/waste.xlsx"
wb = openpyxl.load_workbook(excel_file) # 读取excel文件
sheet = wb.worksheets[0]#读取第一个表
col = sheet["B"]#读取B列
for index,cell in enumerate(col):
if cell.value == name:
result["sort"] = 1
result["name"] = name
result["result"] = cell.value
result["id"] = sheet.cell(index,1).value
result["sortId"] = sheet.cell(index,4).value
print(cell.value,1)
else:
if name in cell.value:
sort = len(name)/len(cell.value)
print(cell.value,sort)
if sort > result["sort"]:
result["sort"] = sort
result["name"] = name
result["result"] = cell.value
result["id"] = sheet.cell(index,1).value
result["sortId"] = sheet.cell(index,4).value
print(cell.value,sort)
if result["sort"] == 0:
print("未识别到是什么垃圾")
return waste,0
else:
print("====识别结果=======")
print(result["name"],"的识别结果是：",result["result"],"是",result["sortId"],"可信度为：",result["sort"])
return result["sortId"],result["sort"]
"""
在图像上标记中文
输入：图片（cv2格式），文字，写到图片上的位置(x,y)，文字颜色，文字大小
输出：图片
"""
def cv2AddChineseText(img, text, position=(0,0), textColor=(0, 255, 0), textSize=30):
if (isinstance(img, np.ndarray)): # 判断是否OpenCV图片类型
img = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
# 创建一个可以在给定图像上绘图的对象
draw = ImageDraw.Draw(img)
# 字体的格式，需要下载
fontStyle = ImageFont.truetype(
"simsun/simsun.ttc", textSize, encoding="utf-8")
# 绘制文本
draw.text(position, text, textColor, font=fontStyle)
# 转换回OpenCV格式
return cv2.cvtColor(np.asarray(img), cv2.COLOR_RGB2BGR)
if __name__ == '__main__':
# 识别的图片
# url = '/Users/wangyu/Desktop/shicai.jpg'
# url = '/Users/wangyu/Desktop/fish.jpg'
url = '/Users/wangyu/Desktop/hongzao.jpg'
img = cv2.imread(url)
# 百度账号信息
AccessKey = ''
SecretKey = ''
# 获取小票识别结果
access_token = getToken(AccessKey,SecretKey)
img_result,score = getResult(url,access_token)
if img_result != "未识别该图片":
waste_result,sort = getWaste(img_result)
image = cv2AddChineseText(image, "图片识别："+img_result, (30, 50), (255,0,0), 50)
image = cv2AddChineseText(image, "score: {:.2f}".format(score), (500, 50), (255,0,255), 30)
image = cv2AddChineseText(image, "垃圾分类："+waste_result, (30, 120), (255,0,0), 50)
image = cv2AddChineseText(image, "sort: {:.2f}".format(sort), (500, 120), (255,0,255), 30)
plt.figure(figsize=(5,5))
plt.imshow(cv2.cvtColor(image, cv2.COLOR_BGR2RGB))#BGR转RGB
plt.xlabel(u'image')
plt.show()总结
通过百度API图像接口实现了单物种的图像识别，取识别率最高的一个结果；通过excel匹配，实现垃圾分类提示。垃圾分类匹配规格达2k+。
图像识别目前仅用了一个类别的识别，百度组合API的识别试了一下，暂时不ok，报400的错误，传参这里还有问题，思考中。
不限于API识别图像，也可以使用其他方案识别图像。
本文章为转载内容，我们尊重原作者对文章享有的著作权。如有内容错误或侵权问题，欢迎原作者联系我们进行内容更正或删除文章。
