所需的输出
输入：使用OpenCV或REST摄像机URL的摄像机源。 （此问题无关）
输出：经过一些OpenCV处理后流jpeg图像
到目前为止，我已经基于$destinationFolder = 'PATH OF THE DESTINATION FOLDER'
$sourceFile = 'FULL PATH AND FILENAME OF THE SOURCE FILE'
# split the filename into a basename and an extension variable
$baseName = [System.IO.Path]::GetFileNameWithoutExtension($sourceFile)
$extension = [System.IO.Path]::GetExtension($sourceFile)
# you could also do it like this:
# $fileInfo = Get-Item -Path $sourceFile
# $baseName = $fileInfo.BaseName
# $extension = $fileInfo.Extension
# get an array of all filenames (name only) of the files with a similar name already present in the destination folder
$allFiles = @(Get-ChildItem $destinationFolder -File -Filter "$baseName*$extension" | Select-Object -ExpandProperty Name)
# construct the new filename
$newName = $baseName + $extension
$count = 1
while ($allFiles -contains $newName) {
# add a sequence number in brackets to the end of the basename until it is unique in the destination folder
$newName = "{0}({1}){2}" -f $baseName, $count++, $extension
}
# construct the new full path and filename for the destination of your Copy-Item command
$targetFile = Join-Path -Path $destinationFolder -ChildPath $newName
Copy-Item -Path $sourceFile -Destination $targetFile
tutorial
输入：图片文件作为POST请求
输出：带有图像路径的GET请求端点
Falcon响应如下：
HTTP / 1.1 201已创建
连接：关闭
日期：2018年12月3日星期一13:08:14 GMT
服务器：gunicorn / 19.7.1
内容长度：0
内容类型：application / json; charset = UTF-8
位置：/images/e69a83ee-b369-47c3-8b1c-60ab7bf875ec.jpg
上面的代码有2个问题：
那么，如何读取import mimetypes
import os
import re
import uuid
import cv2
import io
import falcon
from falcon import media
import json
import msgpack
class Collection(object):
def __init__(self, image_store):
self._image_store = image_store
def on_get(self, req, resp):
# TODO: Modify this to return a list of href's based on
# what images are actually available.
doc = {
'images': [
{
'href': '/images/1eaf6ef1-7f2d-4ecc-a8d5-6e8adba7cc0e.png'
}
]
}
resp.data = msgpack.packb(doc, use_bin_type=True)
resp.content_type = falcon.MEDIA_MSGPACK
resp.status = falcon.HTTP_200
def on_post(self, req, resp):
name = self._image_store.save(req.stream, req.content_type)
# Unnecessary Hack to read the saved file in OpenCV
image = cv2.imread("images/" + name)
new_image = do_something_with_image(image)
_ = cv2.imwrite("images/" + name, new_image)
resp.status = falcon.HTTP_201
resp.location = '/images/' + name
class Item(object):
def __init__(self, image_store):
self._image_store = image_store
def on_get(self, req, resp, name):
resp.content_type = mimetypes.guess_type(name)[0]
resp.stream, resp.stream_len = self._image_store.open(name)
class ImageStore(object):
_CHUNK_SIZE_BYTES = 4096
_IMAGE_NAME_PATTERN = re.compile(
'[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}\.[a-z]{2,4}$'
)
def __init__(self, storage_path, uuidgen=uuid.uuid4, fopen=io.open):
self._storage_path = storage_path
self._uuidgen = uuidgen
self._fopen = fopen
def save(self, image_stream, image_content_type):
ext = mimetypes.guess_extension(image_content_type) # Issue with this code, Not returning the extension so hard coding it in next line
ext = ".jpg"
name = '{uuid}{ext}'.format(uuid=self._uuidgen(), ext=ext)
image_path = os.path.join(self._storage_path, name)
with self._fopen(image_path, 'wb') as image_file:
while True:
chunk = image_stream.read(self._CHUNK_SIZE_BYTES)
if not chunk:
break
image_file.write(chunk)
return name
def open(self, name):
# Always validate untrusted input!
if not self._IMAGE_NAME_PATTERN.match(name):
raise IOError('File not found')
image_path = os.path.join(self._storage_path, name)
stream = self._fopen(image_path, 'rb')
stream_len = os.path.getsize(image_path)
return stream, stream_len
def create_app(image_store):
api = falcon.API()
api.add_route('/images', Collection(image_store))
api.add_route('/images/{name}', Item(image_store))
api.add_sink(handle_404, '')
return api
def get_app():
storage_path = os.environ.get('LOOK_STORAGE_PATH', './images')
image_store = ImageStore(storage_path)
return create_app(image_store)
数据作为numpy数组？更重要的是，我需要进行哪些更改才能从此服务流式传输图像？
P.S。很长的歉意
答案 0 :(得分：1)
我找到了一个很好用的解决方案。有关更多信息，请查看this beautiful code。
learning_curves(estimator=RFReg, X=X_train, y=y_size, train_sizes= train_sizes)