process_image

This method uses the EyeOnIt Large Vision Model to process a single base64-encoded image.

Method: POST

Endpoint: /process_image

Body

{
    "file": "<base64 encoded image string>",
    "return_image": true,
    "regions": [
        {
            "enabled": true,
            "name": "Back Alley",
            "polygon": [
                {
                    "x": 528,
                    "y": 175
                },
                {
                    "x": 876,
                    "y": 166
                },
                {
                    "x": 896,
                    "y": 285
                },
                {
                    "x": 506,
                    "y": 279
                },
                {
                    "x": 528,
                    "y": 175
                }
            ],
            "detection_configs": [
                {
                    "class_name": "vehicle",
                    "class_threshold": 10,
                    "object_size": 150,
                    "object_descriptions": [
                        {
                            "text": "black sedan",
                            "threshold": 90,
                            "alert": true,
                            "background_prompt": false
                        },
                        {
                            "text": "vehicle",
                            "threshold": 90,
                            "alert": false,
                            "background_prompt": false
                        },
                        {
                            "text": "street building",
                            "threshold": null,
                            "alert": false,
                            "background_prompt": true
                        }
                    ],
                    "conditions": [
                        {
                            "type": "count_greater_than",
                            "count": 0,
                            "line_name": null,
                            "alert_direction": null
                        }
                    ],
                    "alert_seconds": 0.3,
                    "reset_seconds": 2.0,
                    "vms_config": null
                }
            ]
        },
    ],
}

Body Attribute Descriptions

Name	Type	Description
file	string	the base64 encoding of the image to process
return_image	bool	true to return an annotated image. If false, no image will be returned.
regions	Region[]	the array of regions to use for processing

Response

{
    "success": true,
    "message": null,
    "data": {
        "detections": [
        ],
    }
}

Response Attribute Descriptions

Name	Type	Description
success	boolean	true if the operation was successful. false otherwise.
message	string	a human readable message describing the operation result
data	JSON	provides a JSON array of data for detected objects as described below

Data Object Attributes

The data element will contain a "detections" array. The objects in the array will take one of two forms. If the detection configuration that trigged the detection included a common object type, the data element will be a Detected Object as shown below:

{
    "success": true,
    "message": null,
    "data": {
        "detections": [
            {
                "class_name": "person",
                "region": "Back Alley",
                "objects": [
                    {
                        "object_description": "person holding weapon",
                        "confidence": 83,
                        "bounds": {
                            "left": 380,
                            "top": 105,
                            "width": 186,
                            "height": 522
                        }
                    },
                    {
                        "object_description": "person",
                        "confidence": 93,
                        "bounds": {
                            "left": 815,
                            "top": 123,
                            "width": 74,
                            "height": 236
                        }
                    }
                ]
            }
        ],
        "image": "<base64-encoded image>"
    }
}

If the detection configuration that trigged the detection did not include a common object type, the element in the detections array will have the following form:

{
    "success": true,
    "message": null,
    "data": {
        "detections": [
            {
                "region": "Back Alley",
                "object_descriptions": [
                    {
                        "text": "person",
                        "confidence": 67
                    },
                    {
                        "text": "person holding weapon",
                        "confidence": 99
                    }
                ]
            }
        ],
        "image": "<base64-encoded image>"
    }
}