Check visual ML api

[mauricebrinkmann]

curl commands to

• train classification
• predict

[DanielWunder]

using Microsoft Costum Vision API you need more than just one http request to train a classification: 1st request: Create a project (if not using an existing project) 2nd request: Create a tag/classification 3rd request: Send images with tag 4th request: Train project

[DanielWunder]

https://southcentralus.dev.cognitive.microsoft.com/docs/services/Custom_Vision_Training_3.0/operations/5c771cdcbf6a2b18a0c3b80f

[DanielWunder]

curl -X POST \ 'https://{region}.api.cognitive.microsoft.com/customvision/v3.0/training/projects/{project-id}/images?tagIds={tagIds}' \ -H 'Content-Type: multipart/form-data' \ -H 'Trainig-Key: {subcription-key}' \ -H 'Training-key: {subcription-key}' \ -H 'cache-control: no-cache' \ -H 'content-type: multipart/form-data; boundary=----WebKitFormBoundary7MA4YWxkTrZu0gW' \ -F '=@{path1}' \ -F '=@{path2}'

... add up to 64 files

[DanielWunder]

1st request: https://southcentralus.dev.cognitive.microsoft.com/docs/services/Custom_Vision_Training_3.0/operations/5c771cdcbf6a2b18a0c3b80f 2nd request: https://southcentralus.dev.cognitive.microsoft.com/docs/services/Custom_Vision_Training_3.0/operations/5c771cdcbf6a2b18a0c3b81f 3rd request: https://southcentralus.dev.cognitive.microsoft.com/docs/services/Custom_Vision_Training_3.0/operations/5c771cdcbf6a2b18a0c3b7ff (see example above) 4th request: https://southcentralus.dev.cognitive.microsoft.com/docs/services/Custom_Vision_Training_3.0/operations/5c771cdcbf6a2b18a0c3b813

[mauricebrinkmann]

It has been decided not to include this issue in the 0.2 release for the hackathon demo in favor to a stable demo and well-prepared presentation, which includes these ideas as a thoughtful outlook with detailed documentation of possible next steps.