vignettes/Detect_Labels.md

In mgahan/WanderingEye: Extracts imaage data from various cloud providers

Load library

# devtools::install_github("mgahan/WanderingEye")
library(WanderingEye)

Sample files

In order to test the package, it is good to have a set of sample files. Feel free to use your own, but here some sample files that I uploaded to Google Cloud

# Download file
downloadURL <- "https://storage.googleapis.com/mike-public-data/SampleAnimalPhotos.zip"
destFile <- "SampleAnimalPhotos.zip"
if (!file.exists(destFile)) {
  download.file(url=downloadURL, destfile = destFile)
}

# Unzip file
destDir <- gsub(".zip","",destFile)
if (!dir.exists(destDir)) {
  unzip(zipfile=destFile)
}

File explore

We can first check out how many files we have available to us.

FileListDat <- data.table(Filename=list.files(path=destDir, recursive = TRUE, full.names = TRUE))
kable(head(FileListDat))

Filename

SampleAnimalPhotos/12227_tapir.jpg SampleAnimalPhotos/12231_porcupine.jpg SampleAnimalPhotos/3743-2.JPG SampleAnimalPhotos/ave rapaz-3734-19.JPG SampleAnimalPhotos/Cchinga B-Rancho Chico 3715-14.JPG SampleAnimalPhotos/chunga 3728-5.JPG

Google Cloud Vision

We can now forge ahead and attempt to detect labels of the image using Google Cloud Vision. If you have not gone through the process of extracting the Google Cloud Vision API keys, please do this before moving onto this next step.

SampleImage <- FileListDat[35, Filename]

GoogleOutput <- 
  googleCloudVision(
    imagePath=SampleImage,
    feature = "LABEL_DETECTION", 
    numResults = 10,
    API_KEY = Sys.getenv("GCLOUD_VISION_API_KEY"))

kable(GoogleOutput[])

mid description score /m/083jv white 0.9630147 /m/019sc black 0.9617918 /m/01g6gs black and white 0.9549949 /m/04rky mammal 0.9270029 /m/035qhg fauna 0.9162718 /m/03d49p1 monochrome photography 0.9149839 /m/01280g wildlife 0.9146594 /m/04hgtk head 0.8594266 /m/0898b zebra 0.8483723 /m/05wkw photography 0.8124335

Microsoft Computer Vision

MicrosoftOutput <- 
  microsoftComputerVision(
    imagePath=SampleImage,
    feature = "analyze",
    MICROSOFT_API_ENDPOINT = Sys.getenv("MICROSOFT_API_ENDPOINT"),
    MICROSOFT_API_KEY1 = Sys.getenv("MICROSOFT_API_KEY1"))

kable(MicrosoftOutput$Tags[])

name confidence hint zebra 0.9985831 animal animal 0.9855506 NA mammal 0.7482966 animal

kable(MicrosoftOutput$Descriptions[])

text confidence a zebra is looking at the camera 0.8817827

kable(MicrosoftOutput$Meta[])

width height format 1280 1024 Jpeg

AWS Rekognition

AWSOutput <- 
  awsRekognition(
    imagePath=SampleImage,
    targetPath = NULL, 
    feature = "detect-labels",
    AWS_ACCESS_KEY_ID = Sys.getenv("AWS_ACCESS_KEY_ID"),
    AWS_SECRET_ACCESS_KEY = Sys.getenv("AWS_SECRET_ACCESS_KEY"),
    AWS_BUCKET = Sys.getenv("AWS_BUCKET"),
    AWS_DEFAULT_REGION = Sys.getenv("AWS_DEFAULT_REGION"))

kable(AWSOutput[])

Feature Score Description File LABELS 96.46909 Animal SampleAnimalPhotos/IMG_0095.JPG LABELS 96.46909 Mammal SampleAnimalPhotos/IMG_0095.JPG LABELS 96.46909 Zebra SampleAnimalPhotos/IMG_0095.JPG LABELS 84.84165 Ct Scan SampleAnimalPhotos/IMG_0095.JPG LABELS 84.84165 X-Ray SampleAnimalPhotos/IMG_0095.JPG

IBM Watson Visual Recognition

IBMOutput <-IBMWatsonVision(
  imagePath=SampleImage, 
  IBM_WATSON_API = Sys.getenv("IBM_WATSON_API"),
  IBM_WATSON_VERSION = Sys.getenv("IBM_WATSON_VERSION"))

kable(IBMOutput[])

Class Score Hierarchy zebra 0.893 /animal/mammal/odd-toed ungulate (hoofed mammal)/zebra odd-toed ungulate (hoofed mammal) 0.992 NA mammal 0.992 NA animal 0.992 NA grevy's zebra 0.803 /animal/mammal/odd-toed ungulate (hoofed mammal)/grevy's zebra mountain zebra 0.5 /animal/mammal/odd-toed ungulate (hoofed mammal)/mountain zebra coal black color 0.961 NA black color 0.893 NA

ClarfAI

ClarifaiOutput <- clarifaiPredict(
  imagePath=SampleImage, 
  CLARIFAI_API_KEY = Sys.getenv("CLARIFAI_API_KEY"))

kable(ClarifaiOutput[])

id name value app_id File ai_786Zr311 no person 0.9927144 main SampleAnimalPhotos/IMG_0095.JPG ai_RmpTltl9 stripe 0.9916375 main SampleAnimalPhotos/IMG_0095.JPG ai_8Z5lBHrh zebra 0.9911756 main SampleAnimalPhotos/IMG_0095.JPG ai_fbqvMwRm wildlife 0.9887770 main SampleAnimalPhotos/IMG_0095.JPG ai_tBcWlsCp nature 0.9869842 main SampleAnimalPhotos/IMG_0095.JPG ai_vfV1Zf9w horizontal 0.9857936 main SampleAnimalPhotos/IMG_0095.JPG ai_SzsXMB1w animal 0.9585370 main SampleAnimalPhotos/IMG_0095.JPG ai_5DDJGZxT skin 0.9446589 main SampleAnimalPhotos/IMG_0095.JPG ai_Zmhsv0Ch outdoors 0.9396830 main SampleAnimalPhotos/IMG_0095.JPG ai_XNzGRk0F side view 0.9283558 main SampleAnimalPhotos/IMG_0095.JPG ai_T85WqSNl camouflage 0.9162262 main SampleAnimalPhotos/IMG_0095.JPG ai_43sQsmXM safari 0.8985972 main SampleAnimalPhotos/IMG_0095.JPG ai_bBl132T0 zoo 0.8980871 main SampleAnimalPhotos/IMG_0095.JPG ai_6pRrC0WT danger 0.8656015 main SampleAnimalPhotos/IMG_0095.JPG ai_j6rltf8j elegant 0.8605607 main SampleAnimalPhotos/IMG_0095.JPG ai_xDm4LRvF zoology 0.8495933 main SampleAnimalPhotos/IMG_0095.JPG ai_SVshtN54 one 0.8491527 main SampleAnimalPhotos/IMG_0095.JPG ai_N6BnC4br mammal 0.8446153 main SampleAnimalPhotos/IMG_0095.JPG ai_0Ngv01Hf contrast 0.8444433 main SampleAnimalPhotos/IMG_0095.JPG ai_wLQ7hLvK identity 0.8337676 main SampleAnimalPhotos/IMG_0095.JPG

mgahan/WanderingEye documentation built on May 22, 2019, 12:38 p.m.