Meetup Paris Cognitives Services #2 - Azure Vision

Quick presentation of Azure Vision capabilities

Transcript

1. Cognitives Services Paris #2 Sylvain Pontoreau At ZENIKA
2. Premier Field Engineer Apps Domain
4. Microsoft's vision reset comes after CEO of Google began saying that the world is shifting from being mobile-first to AI-first.
7. COMPUTER VISION CONTENT MODERATOR FACE EMOTION VIDEO INDEXER CUSTOM VISION
8. "a cat lying on a bed" with probability of 87% Description - "indoor" with probability of 94% - "cat" with probability of 94% Tags - "animal_cat" with probability of 87% Category
9. - Our mission - Empower every person - and every organization - on the planet to - achieve more. Text - HELLO COGNITIVES - SERVICES - PARIS Text
10. JSON { "boundingBox": [1112, 597, 1360, 596, 1361, 614, 1113, 615], "text": "Dossier UWUTWL", "words": [{ "boundingBox": [1115, 594, 1221, 596, 1218, 617, 1112, 615], "text": "Dossier" }, { "boundingBox": [1242, 596, 1364, 598, 1361, 620, 1239, 617], "text": "UWUTWL" }] }
11. "boundingBox": [1112, 597, 1360, 596, 1361, 614, 1113, 615]
12. https://westeurope.api.cognitive.microsoft.com/vision/v1.0/ocr URL - language - orientation detection Parameters URL / Stream Content Subscription key Header
13. male Gender 53 Age No Makeup Happiness Emotion No Beard Yes Smile Blond & Gray Hair color x=509, y=714 Nose tip
14. VS Probability 19%
15. VS Probability 79%
16. Detection Attributes Verification Similar Group ... Location and size Pose, Gender, Age, Hair … Authentication, Identity Match faces Goup by similarity
17. Emotion Probabilities Surprise "scores": { "anger": 9.29041E-06, "contempt": 0.000118981574, "disgust": 3.15619363E-05, "fear": 0.000589638, "happiness": 0.06630674, "neutral": 0.00555004273, "sadness": 7.44669524E-06, "surprise": 0.9273863 }
18. var client = new HttpClient(); client.DefaultRequestHeaders.Add("Ocp-Apim-Subscription-Key", "myKey"); string uri = "https://westeurope.api.cognitive.microsoft.com/emotion/v1.0/recognize?"; using (var content = new ByteArrayContent(myImageAsByteArray)) { content.Headers.ContentType = new MediaTypeHeaderValue("application/octet-stream"); var response = await client.PostAsync(uri, content); var json = await response.Content.ReadAsStringAsync(); }
20. 1 1 1 1 1 2 2 2 2 2 2 ? =
21. Recall Precision k-fold cross validation
22. var client = new HttpClient(); client.DefaultRequestHeaders.Add("Prediction-Key", "myKey"); string uri = "https://southcentralus.api.cognitive.microsoft.com/customvision/v1.0/Prediction/{model_id}/image"; using (var content = new ByteArrayContent(myImageAsByteArray)) { content.Headers.ContentType = new MediaTypeHeaderValue("application/octet-stream"); var response = await client.PostAsync(uri, content); var prediction = await response.Content.ReadAsStringAsync(); }
24. https://microsoft.github.io/ELL/ Embedded Learning Library + =
25. https://azure.microsoft.com/en-us/services/cognitive-services/ Documentation https://azure.microsoft.com/en-us/try/cognitive-services/ Try it free