The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.
An R frontend to Azure Computer Vision and Azure Custom Vision, building on the low-level functionality provided by the AzureCognitive package. These services let you leverage the cloud to carry out visual recognition tasks using advanced image processing models, without needing powerful hardware of your own.
The primary repo for this package is at
https://github.com/Azure/AzureVision; please submit issues and PRs
there. It is also mirrored at the Cloudyr org at
https://github.com/cloudyr/AzureVision. You can install the development
version of the package with
devtools::install_github("Azure/AzureVision")
.
To communicate with the Computer Vision service, call the
computervision_endpoint
function with the service URL and
key. Rather than a key, you can also supply an OAuth token obtained with
the AzureAuth package.
library(AzureVision)
<- computervision_endpoint(
vis url="https://accountname.cognitiveservices.azure.com/",
key="account_key"
)
# images can be specified as a filename, Internet URL, or raw vector
<- "https://news.microsoft.com/uploads/2014/09/billg1_print.jpg"
bill_url analyze(vis, bill_url)
# $categories
# name score
# 1 people_ 0.953125
describe(vis, bill_url)
# $tags
# [1] "person" "man" "suit" "clothing" "necktie" "wearing" "glasses" "looking" "holding" "standing" "older"
# [12] "posing" "business" "old" "dressed" "front" "sitting" "black" "hat" "white" "sign" "phone"
# $captions
# text confidence
# 1 Bill Gates wearing a suit and tie 0.9954072
Custom Vision defines two different types of endpoint: a training
endpoint, and a prediction endpoint. To communicate with these, call the
customvision_training_endpoint
and
customvision_prediction_endpoint
functions with the service
URL and key.
# training a model
<- customvision_training_endpoint(
cusvis url="https://location.api.cognitive.microsoft.com/",
key="training_key"
)
# different projects can exist on the one endpoint
list_projects(cusvis)
# create a classification project (one tag/label per image)
<- create_classification_project(cusvis, "myproject")
proj
<- dir("path/to/images/tag1", full.names=TRUE)
img1 <- dir("path/to/images/tag2", full.names=TRUE)
img2 add_images(proj, img1, tags="tag1")
add_images(proj, img2, tags="tag2")
# train the model
<- train_model(proj)
mod
# publish to the prediction resource (use AzureRMR framework to get resource ID)
<- AzureRMR::get_azure_login("mytenant")$
pred_res get_subscription("sub_id")$
get_resource_group("rgname")$
get_cognitive_service("cusvis_prediction")
publish_model(mod, "mymodel", pred_res)
# get predictions from the prediction endpoint
<- customvision_prediction_endpoint(
cusvis_pred url="https://location.api.cognitive.microsoft.com/",
key="prediction_key"
)
# must supply project ID to access the published service
<- proj$project$id
project_id <- classification_service(cusvis_pred, project_id, "mymodel")
cusvis_service
predict(cusvis_service, "testimage.jpg")
You can create Computer Vision and Custom Vision resources using the AzureRMR framework.
For Computer Vision, the available service tiers are F0
(free, limited to 20 API calls per minute and 5k calls per month) and
S1
(up to 10 calls per second). For Custom Vision, the
tiers are F0
(free, limited to 2 projects for training and
10k transactions/month for prediction) and S0
. Note that
Custom Vision requires at least two resources: one for
training, and the other for prediction.
<- AzureRMR::get_azure_login("yourtenant")$
rg get_subscription("sub_id")$
get_resource_group("rgname")
# Computer Vision
$create_cognitive_service("myvis", service_type="ComputerVision", service_tier="S1")
rg
# Custom Vision (training and prediction)
$create_cognitive_service("mycustvis", service_type="CustomVision", service_tier="S0")
rg$create_cognitive_service("mycustvispred", service_type="CustomVision.Prediction", service_tier="S0") rg
These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.