Hugging Face's logo

Spaces:
noahzhy
/
KR_LPR
Running

App Files
KR_LPR
File size: 4,638 Bytes 
39ee507
 
 
 
 
 
ffef866
39ee507
 
 
 
 
074635e
 
39ee507
 
 
5bfef7f
 
39ee507
 
5bfef7f
 
 
 
 
 
 
 
 
 
39ee507
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5bfef7f
 
 
 
 
 
 
 
cb30e70
5bfef7f
 
 
 
cb30e70
5bfef7f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ffef866
cb30e70
 
 
 
 
 
 
 
 
 
5bfef7f
 
cb30e70
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
 
import os
import glob
from itertools import groupby

import cv2
import numpy as np
import gradio as gr
import tensorflow as tf


def get_sample_images():
    list_ = glob.glob(os.path.join(os.path.dirname(__file__), 'samples/*.jpg'))
    # sort by name
    list_.sort(key=lambda x: int(x.split('/')[-1].split('.')[0]))
    return [[i] for i in list_]


def cv2_imread(path):
    return cv2.imdecode(np.fromfile(path, dtype=np.uint8), cv2.IMREAD_UNCHANGED)


def decode_label(mat, chars) -> str:
    # mat is the output of model
    # get char indices along best path
    best_path_indices = np.argmax(mat[0], axis=-1)
    # collapse best path (using itertools.groupby), map to chars, join char list to string
    best_chars_collapsed = [chars[k] for k, _ in groupby(best_path_indices) if k != len(chars)]
    res = ''.join(best_chars_collapsed)
    # remove space and '_'
    res = res.replace(' ', '').replace('_', '')
    return res


def center_fit(img, w, h, inter=cv2.INTER_NEAREST, top_left=True):
    # get img shape
    img_h, img_w = img.shape[:2]
    # get ratio
    ratio = min(w / img_w, h / img_h)

    if len(img.shape) == 3:
        inter = cv2.INTER_AREA
    # resize img
    img = cv2.resize(img, (int(img_w * ratio), int(img_h * ratio)), interpolation=inter)
    # get new img shape
    img_h, img_w = img.shape[:2]
    # get start point
    start_w = (w - img_w) // 2
    start_h = (h - img_h) // 2

    if top_left:
        start_w = 0
        start_h = 0

    if len(img.shape) == 2:
        # create new img
        new_img = np.zeros((h, w), dtype=np.uint8)
        new_img[start_h:start_h+img_h, start_w:start_w+img_w] = img
    else:
        new_img = np.zeros((h, w, 3), dtype=np.uint8)
        new_img[start_h:start_h+img_h, start_w:start_w+img_w, :] = img

    return new_img


def load_dict(dict_path='label.names'):
    with open(dict_path, 'r', encoding='utf-8') as f:
        dict = f.read().splitlines()
    dict = {i: dict[i] for i in range(len(dict))}
    return dict


class TFliteDemo:
    def __init__(self, model_path, blank=85, conf_mode="mean"):
        self.blank = blank
        self.conf_mode = conf_mode
        self.interpreter = tf.lite.Interpreter(model_path=model_path)
        self.interpreter.allocate_tensors()
        self.input_details = self.interpreter.get_input_details()
        self.output_details = self.interpreter.get_output_details()

    def inference(self, x):
        self.interpreter.set_tensor(self.input_details[0]['index'], x)
        self.interpreter.invoke()
        return self.interpreter.get_tensor(self.output_details[0]['index'])

    def preprocess(self, img):
        if isinstance(img, str):
            image = cv2_imread(img)
        else:
            image = img
        image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
        image = center_fit(image, 128, 64, top_left=True)
        image = np.reshape(image, (1, *image.shape, 1)).astype(np.uint8)
        return image

    def get_confidence(self, pred, mode="mean"):
        conf = []
        idxs = np.argmax(pred, axis=-1)
        values = np.max(pred, axis=-1)

        for idx, c in zip(idxs, values):
            if idx == self.blank: continue
            conf.append(c/255)

        if mode == "min":
            return np.min(conf)

        return np.mean(conf)

    def postprocess(self, pred):
        label = decode_label(pred, load_dict())
        conf = self.get_confidence(pred[0], mode=self.conf_mode)
        # keep 4 decimal places
        conf = float('{:.4f}'.format(conf))
        return label, conf


def inference(img):
    # preprocess
    img = demo.preprocess(img)
    # inference
    pred = demo.inference(img)
    # postprocess
    label, conf = demo.postprocess(pred)
    return label, conf


if __name__ == '__main__':
    _TITLE = '''South Korean License Plate Recognition'''
    _DESCRIPTION = '''
    <div>
    <p style="text-align: center; font-size: 1.3em">This is a demo of South Korean License Plate Recognition.
    <a style="display:inline-block; margin-left: .5em" href='https://github.com/noahzhy/KR_LPR_TF/'><img src='https://img.shields.io/github/stars/noahzhy/KR_LPR_TF?style=social' /></a>
    </p>
    </div>
    '''
    # init model
    demo = TFliteDemo(os.path.join(os.path.dirname(__file__), 'model.tflite'))
    interface = gr.Interface(
        fn=inference,
        inputs="image",
        outputs=[
            gr.Textbox(label="Plate Number", type="text"),
            gr.Textbox(label="Confidence", type="text"),
        ],
        title=_TITLE,
        description=_DESCRIPTION,
        examples=get_sample_images(),
    )
    interface.launch()