rodsnjr / ic-api

Image Catalog - Main API

IC API

Image Catalog API.

Event Examples

Given two images, and a catalog filter for red, an green cars in a parking lot (this will have to be a parent node) The API will generate events in a pipeline action (other services will chain new events).

The request will be like:

{
    "id": "102938102938102938",
    "images": [
        {
            "id": "10938103891",
            "image_key": "parking_lot_1.jpg"
        },
        {  
            "id": "1209381028931290",
            "image_key": "random_image_2.jpg" 
        }
    ],
    "scenes" : [{
      "uid": "194",
      "labels": ["Parking Lot"]
    }],
    "detections" : [{
      "uid": "195",
      "labels": ["Car"],
      "depends_on": "194"
    }],
    "colors": [
        {
          "uid": "196",
          "color": "Red",
          "depends_on": "195"
        },
        {
          "uid": "197",
          "color": "Green",
          "depends_on": "195"
        }
    ]
}

The generated events will be:

[
    {
        "uid": "09129078",
        "catalog_id": "102938102938102938",
        "image_key": "parking_lot_1.jpg",
        "subject": "SCENE_RECOGNITION",
        "filters": ["Parking Lot"],
        "children": [{
          "uid": "195",
          "subject": "OBJECT_DETECTION",
          "filters": ["Car"],
          "children": [
              {
                "uid": "196",
                "subject": "COLOR_RECOGNITION",
                "filters": ["Red"]
              },
              {
                "uid": "197",
                "subject": "COLOR_RECOGNITION",
                "filters": ["Green"]
              }
          ]
        }]
    }, 
    {
        "uid": "13098139",
        "catalog_id": "102938102938102938",
        "image_key": "random_image_2.jpg",
        "subject": "SCENE_RECOGNITION",
        "filters": ["Parking Lot"],
        "children": [
              {
                "uid": "196",
                "subject": "COLOR_RECOGNITION",
                "filter": ["Red"]
              },
              {
                "uid": "197",
                "subject": "COLOR_RECOGNITION",
                "filter": ["Green"]
              }
          ]
        }]
    }
]

The AI Service will then chain one new event for the positive filter.

[
    {
        "uid": "29083102938",
        "catalog_id": "102938102938102938",
        "image_key": "parking_lot_1.jpg",
        "subjects": "OBJECT_DETECTION",
        "filters": ["Car"],
        "children": [
              {
                "uid": "196",
                "subject": "COLOR_RECOGNITION",
                "filters": ["Red"]
              },
              {
                "uid": "197",
                "subject": "COLOR_RECOGNITION",
                "filters": ["Green"]
              }
        ]
    }
]

Then the service will chain for each car in the picture a new event for the given task. And also this event might be specialized and will contain specific metadata (such as object positions). Let's say there are two cars in the picture, the next event will be:

[
    {
        "uid": "019381908319038",
        "catalog_id": "102938102938102938",
        "image_key": "parking_lot_1.jpg",
        "subjects": "COLOR_RECOGNITION",
        "filters": ["Red"],
        "metadata": {
          "object": {
            "x": "223",
            "y": "255",
            "w": "100",
            "h": "250"
          }
        }
    },
    { 
        "uid": "8967464568735489",
        "catalog_id": "102938102938102938",
	    "image_key": "parking_lot_1.jpg",
	    "subjects": ["COLOR_RECOGNITION"],
	    "filters": ["Red"],
        "metadata": {
          "object": {
            "x": "430",
            "y": "380",
            "w": "225",
            "h": "365"
          }
        }
    },
    {
        "uid": "8435498431894351894",
        "catalog_id": "102938102938102938",
        "image_key": "parking_lot_1.jpg",
        "subjects": "COLOR_RECOGNITION",
        "filters": ["Green"],
        "metadata": {
          "object": {
            "x": "223",
            "y": "255",
            "w": "100",
            "h": "250"
          }
        }
    },
    { 
        "uid": "6846543219684321918987",
        "catalog_id": "102938102938102938",
	    "image_key": "parking_lot_1.jpg",
	    "subjects": ["COLOR_RECOGNITION"],
	    "filters": ["Green"],
        "metadata": {
          "object": {
            "x": "430",
            "y": "380",
            "w": "225",
            "h": "365"
          }
        }
    }
]

Lastly this service will chain the result event. TODO

ic-api's Issues

Create the API Layer

Still have to create the API Layer

• Request/Response Objects
• API Validation Layer
• Decouple API functions from the service layer
• Unit Tests
• Update the Behavior Tests
• Tests in the Quart App

Test Coverage in Service

Coverage for the Service Layer

• Image Layer
• Catalog Layer
• Event Layer

Adjust the Pytests

Including

• Coverage
• Make it run properly
• Avoid running unwanted things (like configurations, or providers, or other things)

Will have to probably redesign the project modules to make them more SOLID