AutoVMAF API

API to create, run and monitor autovmaf jobs.

Current features:

• Possible to start new autoabr jobs.
• Possible to run multiple autoabr jobs in parallel.
• Monitor how long a job have been running (milliseconds) and the status i.e ACTIVE/INACTIVE.
• Download and compile VMAF results from S3 for all measured bitrates and resolutions for a specific job.

New Autoabr workers will automatically be created if all current ones are in use. This makes it really easy to create and run jobs in parallel.

Setup

Environment

The following environment variables need to be set.

AWS_ACCESS_KEY_ID=MyAccessKeyID
AWS_SECRET_ACCESS_KEY=MySecretAccessKey
AWS_REGION=eu-north-1
LOAD_CREDENTIALS_FROM_ENV=true // If false, it will load credentials from ~/.aws/credentials

Using an .env file is supported. Just rename .env.example to .env and insert your values.

Initialize

To initialize a new AutoabrService do:

import { AutoabrService } from "@eyevinn/autoabr-api";

// initialize a new instance of AutoabrService
const autoabrService = new AutoabrService();
// register the routes
autoabrService.listen(3000);

The Autoabr service is now up and running and available on port 3000.

Endpoints

Available endpoints are:

Endpoint	Method	Description
/	GET	Heartbeat endpoint of service
/autoabr	POST	Create a new autoabr job. Provide JSON body with settings
/autoabr	GET	List all active autoabr workers
/autoabr/:id	GET	List info about a specific autoabr worker
/autoabr/result/:output/	GET	Download VMAF results from S3 and compile it into JSON. NOTE: Trailing slash is not optional!
/autoabr/result/:output/:model	GET	Download VMAF results from S3 and compile it into JSON for a specific model
?format=csv or ?format=tsv	-	Optional query parameter for the /autoabr/result/ enpoints. Compiles results into CSV or TSV.
/autoabr/cache	DELETE	Clear MediaConvert and AWS pipeline settings cache

Example requests

Start a job

To start a new Autoabr job send a POST request to the /autoabr endpoint with :

{
  "encodingSettingsUrl": "s3://bucket-name/encoding-profile.json",
  "pipelineUrl": "s3://bucket-name/pipeline.json",
  "job": {
    "name": "job-name",
    "reference": "s3://bucket-name/reference.mov",
    "models": ["UHD"],
    "bitrates": [10000000, 12800000],
    "resolutions": [
      {
        "width": 768,
        "height": 432
      },
      {
        "width": 1280,
        "height": 720
      },
      {
        "width": 1920,
        "height": 1080
      },
      {
        "width": 2560,
        "height": 1440
      },
      {
        "width": 3840,
        "height": 2160
      }
    ]
  }
}

If the pipelineUrl and encodingSettingsUrl haven't been set it will use the default settings that can be found in src/resources.

Get results

The endpoint /autoabr/result/:output/ (output is the output specified in the job payload) will download and process all resulting VMAF files from AWS and return the result. This process can take a while depending on how many resolutions and bitrates that have been measured. This means that the response from the endpoint can take several seconds.

Example JSON Response

GET /autoabr/result/job-name/

{
  "id": "BxTH45aRiyAAq_TBbbHqH",
  "status": "INACTIVE",
  "result": {
    "job-name": {
      "PhoneHD": {},
      "HD": {
        "1280x720_10000000_vmaf.json": 91.12216,
        "1280x720_12800000_vmaf.json": 91.12216,
        "1920x1080_10000000_vmaf.json": 97.427916,
        "1920x1080_12800000_vmaf.json": 97.427916
      },
      "UHD": {}
    }
  }
}

Example CSV Response

GET /autoabr/result/job-name/?format=csv

jobname,model,width,height,bitrate,score
job-name,HD,1280,720,10000000,91.12216
job-name,HD,1280,720,12800000,91.12216
job-name,HD,1920,1080,10000000,97.427916
job-name,HD,1920,1080,12800000,97.427916

Example TSV Response

GET /autoabr/result/job-name/?format=tsv

jobname	model	width	height	bitrate	score
job-name	HD	1280	720	10000000	91.12216
job-name	HD	1280	720	12800000	91.12216
job-name	HD	1920	1080	10000000	97.427916
job-name	HD	1920	1080	12800000	97.427916