license_plates_recognition

Update 22_07_17

• test with text detection:
  • no detection: 0.0580617875394309 s/img
  • has detection: 0.07414934519306778 s/img
  • ->deviant: 0.016089 s/img
  • -> USE DETECTION

Update 22_05_07

• Change port to localhost:8400
• Test with only recognition model paddle.
• Inference time with 1 plate:
  • Only recognition model: ~0.05s
  • Use PaddleOCR interface: ~0.07s

Update 22_04_29

• crop plate into 2 parts -> ocr time decreases 0.25s -> 0.05~0.1s (because ocr model doesn't detect text)
• OCR Time: 0.05 ~ 0.1s
• Total time in one image: ~0.2s
• Time by call API to server: ~0.25s (client and server in the same device , i5 8250, run on cpu)

Update 22_04_28

• add api, default run on localhost:8100
• add function to test on video. Change path of video and see the result

Update 22_04_27

• Plate detection from openvino, inference time: 0.05 second/ image
• Plate OCR from PaddleOCR, inference time (including plate detection): 0.28 second/ image

Fail case:

• B -> 3: because glare

API for C#

Use: python app.py
Url: http://127.0.0.1:8400/plate/

var client = new RestClient("http://0.0.0.0:8400/plate/");
client.Timeout = -1;
var request = new RestRequest(Method.POST);
request.AddParameter("name_cam", "");
request.AddFile("image", "path_of_image");
IRestResponse response = client.Execute(request);
Console.WriteLine(response.Content);

Input:

• name_cam: str
• image: file

Result Format

• Success

  {
    "code": 200,
    "plate_text": "12-B3 456.78",
    "msg": "success"
  }
  

• Fail

  {
    "code": 201,   # error
    "error_code": 0,
    "msg": "error message"
  }