- Ruby 2.3+
gem 'scnnr'
gem install scnnr
You can pass configuration options as a block to Scnnr::Client.new.
client = Scnnr::Client.new do |config|
config.api_key = 'YOUR API KEY'
config.api_version = 'v1'
config.timeout = 0 # sec
config.logger = Logger.new(STDOUT) # You can specify an alternative logger.
config.logger.level = :info # You can change the default log level.
end
Request image recognition by an image URL.
Refer: POST /v1/remote/recognitions
url = 'https://example.com/dummy.jpg'
recognition = client.recognize_url(url)
# You can override config.timeout.
recognition = client.recognize_url(url, timeout: 10)
Request image recognition by a binary image.
Refer: POST /v1/recognitions
img = File.open('dummy_image_file', 'rb')
recognition = client.recognize_image(img)
# You can use public parameter as well.
recognition = client.recognize_image(img, timeout: 10, public: true)
Recognition class represents the image recognition result from API.
If the recognition processing is completed, you will get Recognition instance whose state is finished.
recognition.finished?
=> true
recognition.to_h
=> {"id"=>"20170829/ed4c674c-7970-4e9c-9b26-1b6076b36b49",
"objects"=>
[{"bounding_box"=>{"bottom"=>0.2696995, "left"=>0.3842466, "right"=>0.57190025, "top"=>0.14457992},
"category"=>"hat",
"labels"=>
[{"name"=>"ハット", "score"=>0.9985399},
{"name"=>"中折れ", "score"=>0.99334323},
{"name"=>"ストローハット", "score"=>0.95629793},
{"name"=>"ベージュ", "score"=>0.9062561},
{"name"=>"つば広ハット", "score"=>0.7737022},
{"name"=>"ホワイト", "score"=>0.5695046}]},
{"bounding_box"=>{"bottom"=>0.95560884, "left"=>0.41641566, "right"=>0.5212327, "top"=>0.8452401},
"category"=>"shoe",
"labels"=>
[{"name"=>"サンダル", "score"=>0.93934095},
{"name"=>"ホワイト", "score"=>0.74320596},
{"name"=>"パンプス", "score"=>0.70763165},
{"name"=>"サボ", "score"=>0.69153166},
{"name"=>"ストラップ", "score"=>0.66519636},
{"name"=>"ウェッジソール", "score"=>0.6325865},
{"name"=>"オープントゥ", "score"=>0.61965847},
{"name"=>"アンクルストラップ", "score"=>0.576824},
{"name"=>"厚底", "score"=>0.53842664}]},
{"bounding_box"=>{"bottom"=>0.7018228, "left"=>0.35182703, "right"=>0.6113004, "top"=>0.25296396},
"category"=>"dress",
"labels"=>[{"name"=>"グリーン", "score"=>0.9765959}, {"name"=>"ワンピース", "score"=>0.94697183}, {"name"=>"カーキ", "score"=>0.8136864}, {"name"=>"無地", "score"=>0.54719794}, {"name"=>"フレア", "score"=>0.51572186}]}],
"state"=>"finished"}
# If you recognize an image with `public` parameter, the `recognition` has `image`.
recognition.to_h
=> {"id"=>"20170829/ed4c674c-7970-4e9c-9b26-1b6076b36b49",
"image"=>
{"url"=>"some-image-url",
"size"=>{"height"=>400, "width"=>400}},
"objects"=>
[{"bounding_box"=>{"bottom"=>0.2696995, "left"=>0.3842466, "right"=>0.57190025, "top"=>0.14457992},
"category"=>"hat",
"labels"=>
[{"name"=>"ハット", "score"=>0.9985399},
{"name"=>"中折れ", "score"=>0.99334323},
{"name"=>"ストローハット", "score"=>0.95629793},
{"name"=>"ベージュ", "score"=>0.9062561},
{"name"=>"つば広ハット", "score"=>0.7737022},
{"name"=>"ホワイト", "score"=>0.5695046}]},
{"bounding_box"=>{"bottom"=>0.95560884, "left"=>0.41641566, "right"=>0.5212327, "top"=>0.8452401},
"category"=>"shoe",
"labels"=>
[{"name"=>"サンダル", "score"=>0.93934095},
{"name"=>"ホワイト", "score"=>0.74320596},
{"name"=>"パンプス", "score"=>0.70763165},
{"name"=>"サボ", "score"=>0.69153166},
{"name"=>"ストラップ", "score"=>0.66519636},
{"name"=>"ウェッジソール", "score"=>0.6325865},
{"name"=>"オープントゥ", "score"=>0.61965847},
{"name"=>"アンクルストラップ", "score"=>0.576824},
{"name"=>"厚底", "score"=>0.53842664}]},
{"bounding_box"=>{"bottom"=>0.7018228, "left"=>0.35182703, "right"=>0.6113004, "top"=>0.25296396},
"category"=>"dress",
"labels"=>[{"name"=>"グリーン", "score"=>0.9765959}, {"name"=>"ワンピース", "score"=>0.94697183}, {"name"=>"カーキ", "score"=>0.8136864}, {"name"=>"無地", "score"=>0.54719794}, {"name"=>"フレア", "score"=>0.51572186}]}],
"state"=>"finished"}
If the timeout value is zero or nil, you will get Recognition instance whose state is queued.
Then you can fetch the recognition result using Scnnr::Client#fetch.
Refer: GET /v1/recognitions/*
recognition.queued?
=> true
recognition.to_h
=> {"id"=>"20170829/ed4c674c-7970-4e9c-9b26-1b6076b36b49", "state"=>"queued"}
recognition = client.fetch(recognition.id)
recognition.finished?
=> true
Request fashion coordinates generation.
Refer: POST /v1/coordinates
tastes = { casual: 0.7, girly: 0.3 }
coordinate = client.coordinate('tops', ['グレー', 'パーカー'], tastes, target: 3)
coordinate.to_h
=> {"items"=>
[{"category"=>"tops", "labels"=>[{"name"=>"グレー", "score"=>nil}, {"name"=>"パーカー", "score"=>nil}]},
{"category"=>"shoe",
"labels"=>
[{"name"=>"ネイビー", "score"=>nil},
{"name"=>"スニーカー", "score"=>nil},
{"name"=>"ランニング", "score"=>nil}]},
{"category"=>"dress",
"labels"=>
[{"name"=>"デニム", "score"=>nil},
{"name"=>"ブルー", "score"=>nil},
{"name"=>"サロペット", "score"=>nil},
{"name"=>"オーバーオール", "score"=>nil}]}]}
If the recognition processing is not completed within the timeout time or the recognition failed,
you will get an error with Recognition instance.
begin
url = 'https://example.com/dummy.jpg'
recognition = client.recognize_url(url, timeout: 10)
rescue Scnnr::TimeoutError => e
# You can fetch the result just like when the timeout value is zero or nil.
recognition = client.fetch(e.recognition.id, timeout: 10)
recognition.finished? # => true or false
rescue Scnnr::RecognitionFailed => e
# Failed to recognize the image you requested.
recognition = e.recognition
recognition.error? # => true
STDERR.puts "[ERROR] #{e.title}: #{e.detail} #{e.type}"
rescue Scnnr::RequestFailed => e
# Failed to reserve the recognition.
# This kind of errors have no `#recognition` field.
STDERR.puts "[ERROR] #{e.title}: #{e.detail} #{e.type}"
rescue Scnnr::RecognitionNotFound => e
# Failed to find the recognition with the specified ID.
# This kind of errors have no `#recognition` field.
STDERR.puts "[ERROR] #{e.title}: #{e.detail} #{e.type}"
rescue => e
# Unexpected error.
raise
end