The Thermographic PV Inspection API detects several objects in thermographic pictures:

• PV panels
• PV junction boxes
• Hot spots

The API has two endpoints:

• Analyze Image URL (/api_v1_analyze_image_url) - Image is given by an URL.
• Analyze Image Data (/api_v1_analyze_image_data) - Image is given by data in the request and it is enconded in base 64.

In both cases, the response is in JSON format. This tutorial describes the elements in the response.

• image_analyzed_url - Signed URL to the generated analyzed image. This image is available 3 hours for download and it is automatically deleted after that period.

"image_analyzed_url": "https://storage.googleapis.com/pv_api_images/0ba834e9-3dbc-4a64-a31c-6b093843a044_d.jpg?Expires=1608290478"

• num_detections - Number of detected objects in the image.

"num_detections": 3

• detection_ids - Array of integers. Its length is given by num_detections. Each value is the id of the detected object: Pv panel (1) , Pv junction box (2) or Hot spot (3).

 "detection_ids": [1, 2, 3]

• detection_classes - Array of strings. Its length is given by num_detections. Each value is the name of the detected object: “Pv panel”, “Pv junction box” or “Hot spot”.

 "detection_classes": ["Pv panel", "Pv junction box", "Hot spot"]

• detection_scores - Array of floats. Its length is given by num_detections. Each value is the confidence given by our artificial intelligence to each detected object.

 "detection_scores": [0.9037377834320068, 0.8966286182403564, 0.8696811199188232]

• detection_colors - Array of strings. Its length is given by num_detections. Each value is the HTML color name of the detected object: “Chartreuse”, “Turquoise” or “Yellow”.

 "detection_colors": ["Chartreuse", "Turquoise", "Yellow"]

• detection_boxes - Matrix of floats. Its dimension is num_detections x 4. It defines the bounding box of each detected object by means of 4 values (top, left, bottom, right). Such values are normalized in the [0, 1] interval with respect to the image size (height, width).

 "detection_boxes": [
        [
            0.004028689116239548,
            0.2772405445575714,
            0.11143264174461365,
            0.3454528748989105
        ],
        [
            0.0051384977996349335,
            0.40152040123939514,
            0.10535024106502533,
            0.46812471747398376
        ],
        [
            0.005915585905313492,
            0.14954514801502228,
            0.11493740975856781,
            0.22048313915729523
        ]
]

An example of a whole JSON response would be something like this:

{
     "image_analyzed_url": "https://storage.googleapis.com/pv_api_images/0ba834e9-3dbc-4a64-a31c-6b093843a044_d.jpg?Expires=1608290478",
     "num_detections": 3,
     "detection_ids": [1, 2, 3],
     "detection_classes": ["Pv panel", "Pv junction box", "Hot spot"],
     "detection_scores": [0.9037377834320068, 0.8966286182403564, 0.8696811199188232],
     "detection_colors": ["Chartreuse", "Turquoise", "Yellow"],
     "detection_boxes": [
        [
            0.004028689116239548,
            0.2772405445575714,
            0.11143264174461365,
            0.3454528748989105
        ],
        [
            0.0051384977996349335,
            0.40152040123939514,
            0.10535024106502533,
            0.46812471747398376
        ],
        [
            0.005915585905313492,
            0.14954514801502228,
            0.11493740975856781,
            0.22048313915729523
        ]
     ]
}

This is an example of a generated analyzed image provided by the Thermographic PV Inspection API and given by the image_analyzed_url property in the JSON response.