UltimateALPR - Output JSON Documentation

Jack Heaffey

unread,

Oct 12, 2023, 11:41:08 AM10/12/23

to discuss-doubango

Hi,

Is there any documentation that describes the values in the JSON generated by the process function? Specifically, the output JSON contains a “confidence” field – is this the confidence in the licence plate, or the colour/make/model, or some combination of both?

I have observed that the confidence could be 99% when the JSON does not contain a licence plate, i.e no “text” field.

Here is the output JSON:

{
  "duration": 509,
  "frame_id": 12,
  "plates": [
    {
      "car": {
        "bodyStyle": [
          { "confidence": 94.53125, "klass": 10, "name": "suv" },
          { "confidence": 5.078125, "klass": 11, "name": "van" },
          { "confidence": 0.0, "klass": 0, "name": "bus" },
          { "confidence": 0.0, "klass": 1, "name": "campervan" },
          { "confidence": 0.0, "klass": 2, "name": "convertible" }
        ],
        "color": [
          { "confidence": 98.4375, "klass": 0, "name": "black" },
          { "confidence": 1.5625, "klass": 8, "name": "red" },
          { "confidence": 0.0, "klass": 1, "name": "blue" },
          { "confidence": 0.0, "klass": 2, "name": "brown" },
          { "confidence": 0.0, "klass": 3, "name": "gold" }
        ],
        "confidence": 99.60938,
        "makeModelYear": [
          {
            "confidence": 63.67188,
            "klass": 1445,
            "make": "nissan",
            "model": "pathfinder",
            "year": "0-9"
          },
          {
            "confidence": 15.23438,
            "klass": 1117,
            "make": "land rover",
            "model": "lr4",
            "year": "0-9"
          },
          {
            "confidence": 6.640625,
            "klass": 1116,
            "make": "land rover",
            "model": "lr3",
            "year": "0-9"
          },
          {
            "confidence": 3.515625,
            "klass": 1118,
            "make": "land rover",
            "model": "range rover",
            "year": "0-9"
          },
          {
            "confidence": 1.953125,
            "klass": 1112,
            "make": "land rover",
            "model": "discovery",
            "year": "0-9"
          }
        ],
        "warpedBox": [
          58.15044, 8.311544, 498.0512, 8.311544, 498.0512, 357.3963, 58.15044,
          357.3963
        ]
      }
    }
  ]
}

Mamadou DIOP

unread,

Oct 12, 2023, 7:01:43 PM10/12/23

to doub...@googlegroups.com, Jack Heaffey

Hi,

The right developer group is https://groups.google.com/g/doubango-ai

The 99.60938% confidence is for the car detection, you clearly see that it's on the "car" scope.

"I have observed that the confidence could be 99% when the JSON does not contain a licence plate, i.e no “text” field. " -> that's not possible.

For me the JSON is pretty clear and easy to understand. Each confidence value is scoped.

https://www.doubango.org/SDKs/anpr/docs/Frequently_Asked_Questions_(FAQ).html#could-you-explain-what-each-field-in-the-json-result-mean

--
You received this message because you are subscribed to the Google Groups "discuss-doubango" group.
To unsubscribe from this group and stop receiving emails from it, send an email to doubango+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/doubango/b5237cb7-fd61-482f-b1ab-f97b89864cc0n%40googlegroups.com.

Jack Heaffey

unread,

Oct 13, 2023, 8:32:37 AM10/13/23

to discuss-doubango

Hi Mamadou,

Many thanks for your response.

I would argue that the JSON fields are not explicitly clear. For example if UltimateALPR had detected a motorcycle, the "car" field would contain a "bodyStyle" containing a motorcycle, which is confusing as a motorcycle is not a car.

"I have observed that the confidence could be 99% when the JSON does not contain a licence plate, i.e no “text” field. " -> that's not possible.

It must be possible as the JSON I provided was generated by UltimateALPR. Note that car_noplate_detect_enabled was enabled at the time.

To clarify, the "confidence" field within the "car" field provides the confidence that a vehicle was detected, which could be a car, motorcycle, SUV, bus etc?