search

DoubangoTelecom / ultimateALPR-SDK

World's fastest ANPR / ALPR implementation for CPUs, GPUs, VPUs and NPUs using deep learning (Tensorflow, Tensorflow lite, TensorRT, OpenVX, OpenVINO). Multi-Charset (Latin, Korean, Chinese) & Multi-OS (Jetson, Android, Raspberry Pi, Linux, Windows) & Multi-Arch (ARM, x86).
https://www.doubango.org/webapps/alpr/
Other
593 stars 157 forks source link

Wrong data when roi is empty #243

Closed Nook2007 closed 2 years ago

Nook2007 commented 2 years ago

OS: android Input image: 1920x1080

{
    "frame_id": 184,
    "job_idx": 1,
    "latency": 1,
    "plates": [{
        "car": {
            "confidence": 94.92188,
            "warpedBox": [463.7406, 1151.879, 676.348, 1151.879, 676.348, 1350.656, 463.7406, 1350.656]
        },
        "confidences": [54.86697, 99.60938, 91.27241, 57.06115, 85.34163, 76.95044, 54.86697, 83.03861],
        "text": "M6348*",
        "warpedBox": [580.775, 1257.656, 638.5829, 1257.656, 638.5829, 1287.826, 580.775, 1287.826]
    }]
}

second value in warped box is wrong (1257.656 > 1080) I noticated this happens when:

"detect_roi":[0.0,0.0,0.0,0.0]

(roi = full image)

And i saved input image - and i don't see cars with this plate number

DoubangoTelecom commented 2 years ago

Please attach the image reproducing the issue. On Android the image could be rotated 90° if we detect the device is in portrait which means the image become 1080×1920. For the car it could be false positive.