ramanaditya / Seasons-of-Serverless-Solution-Lovely-Ladoos

Serverless Solution of Predicting Accuracy of your ladoos, using Custom Vision API and Azure Functions

Seasons of Serverless: Solution, Lovely Ladoos

Blogs

• Medium
• dev.to

About

Each year at this time, we prepare something special to help you learn Serverless once and for all with a series of challenges. This year, we decided to feature seven unique challenges, one per week. Come join us!

• Check Original Repo

Solution

The solution creates a Serverless Endpoint which is used to

• List all the images stored on the Azure Blob Storage
• Upload new image on the Azure Blob Storage
• Predict the already uploaded image using Custom Vision API

Pre-requisites

• Create Resource Group
• Create Azure Blob Storage
• Create a Container
• Deploy the Custom Vision API, you can use this datasets
• Publish your prediction

Running the Serverless Functions

Clone the Project

git clone [e-mail]:ramanaditya/Seasons-of-Serverless-Solution-Lovely-Ladoos.git

Install Dependencies

npm install

Export .env variables

export $(grep -v '^#' .env | xargs)

Required Environment Variables are given in the .env.example

Run the function locally

func start

Checking the Working of API endpoint

• Test API using Postman

Endpoints

Base URL: http://localhost:7071

• /api/predict?imageurl=<Image URL>
• /api/blobs

Requests

endpoint	method	parameters	required
/api/predict	GET	imageurl=<ImageURL>	True
/api/blobs	GET
	POST

Response

/api/blobs: GET

[
  {
    "name": "11624351181437143-11108a41.jpg",
    "properties": {
      "createdOn": "2020-11-18T17:19:44.000Z",
      "lastModified": "2020-11-18T17:19:44.000Z",
      "etag": "0x8D88BE624813E21",
      "contentLength": 46337,
      "contentType": "image/jpeg",
      "contentEncoding": "",
      "contentLanguage": "",
      "contentMD5": {
        "type": "Buffer",
        "data": []
      },
      "contentDisposition": "",
      "cacheControl": "",
      "blobType": "BlockBlob",
      "leaseStatus": "unlocked",
      "leaseState": "available",
      "serverEncrypted": true,
      "accessTier": "Hot",
      "accessTierInferred": true,
      "Content-CRC64": ""
    },
    "objectReplicationMetadata": "",
    "OrMetadata": "",
    "imageUrl": "https://<storage-account>.blob.core.windows.net/<container-name>/11624351181437143-11108a41.jpg"
  }
]

/api/blobs: POST

{
  "message": {
    "etag": "\"0x8D88C9867471CEB\"",
    "lastModified": "2020-11-19T14:35:46.000Z",
    "xMsContentCrc64": {
      "type": "Buffer",
      "data": [236, 13, 94, 16, 86, 9, 229, 124]
    },
    "clientRequestId": "####-####-####-####",
    "requestId": "####-####-####-####",
    "version": "2020-02-10",
    "date": "2020-11-19T14:35:46.000Z",
    "isServerEncrypted": true,
    "content-length": "0",
    "server": "Windows-Azure-Blob/1.0 Microsoft-HTTPAPI/2.0"
  },
  "url": "https://<storage-account>.blob.core.windows.net/<container-name>/image (62).jpeg"
}

/api/predict?imageurl=<Image-URL>: GET

[
  {
    "probability": "94.44%",
    "tagId": "def6ab54-27d1-4604-9bd8-f05e8ec7bc13",
    "tagName": "ladoo",
    "tagType": "Regular"
  },
  {
    "probability": "5.30%",
    "tagId": "c0797ccb-bf83-46be-b07b-08d5f60100e1",
    "tagName": "doughnut",
    "tagType": "Regular"
  },
  {
    "probability": "0.26%",
    "tagId": "36c3a169-0f43-4bc7-ba0b-76845a97b607",
    "tagName": "sesame",
    "tagType": "Regular"
  }
]

Resources

• Visual Studio Code
• Azure Account
• Azure Command-Line Interface (CLI)
• Azure Functions Extension
• Custom Vision
• Azure Blob Storage

Contributors

• Aditya Raman
• Rama Soumya Naraparaju