Tensorflow Lite 对象检测 csv

Question

我正在使用 Tensorflow Lite 为我的 Raspberry Pi 制作对象检测软件。我已经在运行 Tensorflow Lite 的情况下正确设置了我的 Pi，正确设置了我的 Tensorflow Lite 模型制作器，并收集并注释了我的图像。现在，我正在努力将这些图像转换为我的模型制作器可以使用的 .csv 格式。

有人有任何提示、教程或建议吗？

这是我当前的 csv 的样子...

filename,width,height,class,xmin,ymin,xmax,ymax
/content/NameTag (1).jpeg,4032,3024,tag,2120,107,2307,181
/content/NameTag (1).jpeg,4032,3024,tag,2203,1602,2299,1668
/content/NameTag (10).jpeg,4032,3024,tag,2242,220,2368,285
/content/NameTag (10).jpeg,4032,3024,tag,2285,1194,2442,1337
/content/NameTag (10).jpeg,4032,3024,tag,2085,2211,2242,2289
/content/NameTag (100).jpeg,4032,3024,tag,838,2576,1133,2842
/content/NameTag (11).jpeg,4032,3024,tag,2112,2163,2268,2259
/content/NameTag (11).jpeg,4032,3024,tag,2312,1194,2455,1311
/content/NameTag (11).jpeg,4032,3024,tag,2268,211,2385,268
/content/NameTag (12).jpeg,4032,3024,tag,2116,2298,2259,2363
/content/NameTag (12).jpeg,4032,3024,tag,2307,1281,2464,1398
/content/NameTag (12).jpeg,4032,3024,tag,2281,298,2394,359
/content/NameTag (13).jpeg,4032,3024,tag,2790,1968,2933,2050
/content/NameTag (13).jpeg,4032,3024,tag,2985,976,3172,1072
/content/NameTag (13).jpeg,4032,3024,tag,2968,33,3133,124
/content/NameTag (14).jpeg,4032,3024,tag,2107,876,2181,946
/content/NameTag (14).jpeg,4032,3024,tag,2251,1711,2329,1772
/content/NameTag (15).jpeg,4032,3024,tag,2085,1576,2159,1637
/content/NameTag (15).jpeg,4032,3024,tag,2168,1781,2238,1842
/content/NameTag (15).jpeg,4032,3024,tag,2099,833,2190,911
/content/NameTag (16).jpeg,4032,3024,tag,2059,815,2133,885
/content/NameTag (16).jpeg,4032,3024,tag,2068,1620,2155,1689
/content/NameTag (16).jpeg,4032,3024,tag,2094,1820,2185,1889
/content/NameTag (17).jpeg,4032,3024,tag,2125,746,2233,833
/content/NameTag (17).jpeg,4032,3024,tag,2116,1542,2220,1620
/content/NameTag (17).jpeg,4032,3024,tag,2168,1846,2246,1920
/content/NameTag (18).jpeg,4032,3024,tag,2142,624,2242,689
/content/NameTag (18).jpeg,4032,3024,tag,2099,1550,2185,1633
/content/NameTag (18).jpeg,4032,3024,tag,2164,1837,2255,1937
/content/NameTag (19).jpeg,4032,3024,tag,2172,1885,2255,1959
/content/NameTag (19).jpeg,4032,3024,tag,2081,1542,2181,1615
/content/NameTag (19).jpeg,4032,3024,tag,2129,594,2246,668
/content/NameTag (2).jpeg,4032,3024,tag,2464,1794,2533,1872
/content/NameTag (2).jpeg,4032,3024,tag,2107,733,2246,811
/content/NameTag (20).jpeg,4032,3024,tag,3290,1076,3403,1142
/content/NameTag (20).jpeg,4032,3024,tag,3446,1389,3546,1437
/content/NameTag (20).jpeg,4032,3024,tag,3294,1833,3403,1915
/content/NameTag (21).jpeg,4032,3024,tag,3233,963,3338,1020
/content/NameTag (21).jpeg,4032,3024,tag,3442,1263,3538,1320
/content/NameTag (21).jpeg,4032,3024,tag,3272,1868,3381,1937
/content/NameTag (22).jpeg,4032,3024,tag,3207,1976,3316,2055
/content/NameTag (22).jpeg,4032,3024,tag,3425,1302,3516,1359
/content/NameTag (22).jpeg,4032,3024,tag,3120,933,3255,1011
/content/NameTag (23).jpeg,4032,3024,tag,1942,2585,2177,2737
/content/NameTag (23).jpeg,4032,3024,tag,1625,520,1868,685
/content/NameTag (23).jpeg,4032,3024,tag,3046,1163,3194,1298
/content/NameTag (24).jpeg,4032,3024,tag,2225,1146,2307,1233
/content/NameTag (24).jpeg,4032,3024,tag,2277,837,2325,902
/content/NameTag (25).jpeg,4032,3024,tag,2564,2785,2712,2855
/content/NameTag (25).jpeg,4032,3024,tag,1999,1498,2107,1555
/content/NameTag (25).jpeg,4032,3024,tag,2233,646,2338,707
/content/NameTag (26).jpeg,4032,3024,tag,2472,2211,2625,2289
/content/NameTag (26).jpeg,4032,3024,tag,1820,1015,1929,1081
/content/NameTag (26).jpeg,4032,3024,tag,2046,102,2199,211
/content/NameTag (27).jpeg,4032,3024,tag,1755,1520,2033,1707
/content/NameTag (28).jpeg,4032,3024,tag,1694,1550,1964,1720
/content/NameTag (29).jpeg,4032,3024,tag,1585,1550,1894,1720
/content/NameTag (3).jpeg,4032,3024,tag,2203,1037,2307,1133
/content/NameTag (3).jpeg,4032,3024,tag,2481,628,2546,685
/content/NameTag (30).jpeg,4032,3024,tag,1599,1376,1899,1481
/content/NameTag (31).jpeg,4032,3024,tag,2064,1455,2216,1589
/content/NameTag (32).jpeg,4032,3024,tag,2094,1889,2229,1963
/content/NameTag (33).jpeg,4032,3024,tag,2042,1550,2199,1655
/content/NameTag (34).jpeg,4032,3024,tag,712,1542,972,1628
/content/NameTag (34).jpeg,4032,3024,tag,1712,1581,2012,1711
/content/NameTag (35).jpeg,4032,3024,tag,1659,1663,1772,1755
/content/NameTag (36).jpeg,4032,3024,tag,1612,1655,1729,1737
/content/NameTag (37).jpeg,4032,3024,tag,1429,750,1603,872
/content/NameTag (37).jpeg,4032,3024,tag,2381,1528,2442,1594
/content/NameTag (38).jpeg,4032,3024,tag,1964,1215,2042,1281
/content/NameTag (39).jpeg,4032,3024,tag,2433,1398,2533,1455
/content/NameTag (4).jpeg,4032,3024,tag,2381,2268,2516,2333
/content/NameTag (40).jpeg,4032,3024,tag,2081,1628,2259,1737
/content/NameTag (41).jpeg,4032,3024,tag,2046,1489,2229,1611
/content/NameTag (42).jpeg,4032,3024,tag,1272,1555,1577,1715
/content/NameTag (43).jpeg,4032,3024,tag,1877,1320,2099,1468
/content/NameTag (44).jpeg,4032,3024,tag,1972,1563,2181,1681
/content/NameTag (45).jpeg,4032,3024,tag,1564,1342,1925,1533
/content/NameTag (46).jpeg,4032,3024,tag,1916,1489,2142,1655
/content/NameTag (47).jpeg,4032,3024,tag,2085,1615,2242,1676
/content/NameTag (48).jpeg,4032,3024,tag,2799,1776,2925,1920
/content/NameTag (49).jpeg,4032,3024,tag,1746,2176,1933,2272
/content/NameTag (49).jpeg,4032,3024,tag,2159,1350,2290,1463
/content/NameTag (49).jpeg,4032,3024,tag,1912,663,2112,794
/content/NameTag (5).jpeg,4032,3024,tag,2385,2215,2520,2294
/content/NameTag (50).jpeg,4032,3024,tag,1942,733,2020,876
/content/NameTag (50).jpeg,4032,3024,tag,2142,1450,2207,1572
/content/NameTag (51).jpeg,4032,3024,tag,1990,1389,2164,1498
/content/NameTag (51).jpeg,4032,3024,tag,2420,907,2542,972
/content/NameTag (52).jpeg,4032,3024,tag,1872,1055,1981,1120
/content/NameTag (52).jpeg,4032,3024,tag,1899,1372,1999,1442
/content/NameTag (53).jpeg,4032,3024,tag,1855,1642,1968,1724
/content/NameTag (53).jpeg,4032,3024,tag,1872,1320,2007,1394
/content/NameTag (54).jpeg,4032,3024,tag,1833,933,2042,1059
/content/NameTag (55).jpeg,4032,3024,tag,1655,1824,1772,1950
/content/NameTag (56).jpeg,4032,3024,tag,2303,1020,2412,1089
/content/NameTag (57).jpeg,4032,3024,tag,1938,215,2094,311
/content/NameTag (57).jpeg,4032,3024,tag,2238,1046,2346,1111
/content/NameTag (58).jpeg,4032,3024,tag,2081,2011,2155,2089
/content/NameTag (59).jpeg,4032,3024,tag,1751,2050,1933,2172
/content/NameTag (59).jpeg,4032,3024,tag,1059,728,1277,946
/content/NameTag (6).jpeg,3024,4032,tag,776,1575,829,1675
/content/NameTag (6).jpeg,3024,4032,tag,2558,1480,2664,1663
/content/NameTag (60).jpeg,4032,3024,tag,1225,2355,1420,2507
/content/NameTag (61).jpeg,4032,3024,tag,2155,163,2368,333
/content/NameTag (62).jpeg,4032,3024,tag,1538,498,1703,711
/content/NameTag (63).jpeg,4032,3024,tag,1572,168,1703,324
/content/NameTag (64).jpeg,4032,3024,tag,1277,1542,1433,1659
/content/NameTag (65).jpeg,4032,3024,tag,1012,2046,1229,2189
/content/NameTag (65).jpeg,4032,3024,tag,1233,1568,1364,1681
/content/NameTag (66).jpeg,4032,3024,tag,1277,1750,1455,1889
/content/NameTag (67).jpeg,4032,3024,tag,1285,902,1638,989
/content/NameTag (68).jpeg,4032,3024,tag,1516,650,1668,763
/content/NameTag (68).jpeg,4032,3024,tag,1607,1468,1746,1533
/content/NameTag (69).jpeg,4032,3024,tag,1533,463,1720,589
/content/NameTag (69).jpeg,4032,3024,tag,1659,1311,1799,1363
/content/NameTag (69).jpeg,4032,3024,tag,2020,1950,2125,2020
/content/NameTag (7).jpeg,4032,3024,tag,2299,885,2425,946
/content/NameTag (7).jpeg,4032,3024,tag,2368,2724,2499,2820
/content/NameTag (70).jpeg,4032,3024,tag,1603,507,1803,624
/content/NameTag (70).jpeg,4032,3024,tag,1746,1342,1877,1389
/content/NameTag (70).jpeg,4032,3024,tag,1907,1676,2025,1750
/content/NameTag (71).jpeg,4032,3024,tag,1451,1815,1690,1963
/content/NameTag (72).jpeg,4032,3024,tag,2816,472,2894,602
/content/NameTag (72).jpeg,4032,3024,tag,1959,1750,2151,1876
/content/NameTag (73).jpeg,4032,3024,tag,2520,1302,2638,1389
/content/NameTag (73).jpeg,4032,3024,tag,1994,1863,2203,2007
/content/NameTag (74).jpeg,4032,3024,tag,1503,2572,1759,2737
/content/NameTag (74).jpeg,4032,3024,tag,2290,81,2485,268
/content/NameTag (74).jpeg,4032,3024,tag,1677,1024,1916,1194
/content/NameTag (75).jpeg,4032,3024,tag,1285,1468,1612,1907
/content/NameTag (75).jpeg,4032,3024,tag,1120,424,1803,868
/content/NameTag (75).jpeg,4032,3024,tag,2238,2420,2816,2915
/content/NameTag (76).jpeg,4032,3024,tag,877,1,1568,524
/content/NameTag (76).jpeg,4032,3024,tag,2807,1772,3307,2146
/content/NameTag (76).jpeg,4032,3024,tag,1072,2585,1333,3024
/content/NameTag (77).jpeg,4032,3024,tag,672,2289,1416,3024
/content/NameTag (78).jpeg,4032,3024,tag,2581,2302,2768,2755
/content/NameTag (78).jpeg,4032,3024,tag,1890,294,2242,1020
/content/NameTag (79).jpeg,4032,3024,tag,1051,733,2168,2702
/content/NameTag (79).jpeg,4032,3024,tag,2212,555,2525,1820
/content/NameTag (79).jpeg,4032,3024,tag,2529,750,2825,1520
/content/NameTag (8).jpeg,4032,3024,tag,2412,1876,2520,1955
/content/NameTag (80).jpeg,4032,3024,tag,377,1211,1677,2207
/content/NameTag (80).jpeg,4032,3024,tag,1559,915,2377,2150
/content/NameTag (80).jpeg,4032,3024,tag,2351,1002,2968,1589
/content/NameTag (81).jpeg,4032,3024,tag,1638,433,1994,715
/content/NameTag (81).jpeg,4032,3024,tag,2029,1598,2355,1811
/content/NameTag (81).jpeg,4032,3024,tag,1372,2528,1694,2750
/content/NameTag (82).jpeg,4032,3024,tag,2007,450,2416,733
/content/NameTag (82).jpeg,4032,3024,tag,2442,1655,2794,1872
/content/NameTag (82).jpeg,4032,3024,tag,1794,2450,2072,2659
/content/NameTag (83).jpeg,4032,3024,tag,1799,1720,1964,1815
/content/NameTag (84).jpeg,4032,3024,tag,1642,950,1781,1055
/content/NameTag (86).jpeg,4032,3024,tag,2003,1811,2133,1902
/content/NameTag (87).jpeg,4032,3024,tag,1725,1311,1812,1611
/content/NameTag (88).jpeg,4032,3024,tag,512,1763,872,2050
/content/NameTag (89).jpeg,4032,3024,tag,1959,133,2138,411
/content/NameTag (89).jpeg,4032,3024,tag,1538,2189,1820,2655
/content/NameTag (9).jpeg,4032,3024,tag,2259,959,2412,1033
/content/NameTag (90).jpeg,4032,3024,tag,303,328,751,981
/content/NameTag (91).jpeg,4032,3024,tag,2333,1059,2781,1302
/content/NameTag (92).jpeg,4032,3024,tag,1194,2437,1255,2637
/content/NameTag (92).jpeg,4032,3024,tag,2694,1315,2877,1420
/content/NameTag (92).jpeg,4032,3024,tag,1151,524,1238,707
/content/NameTag (93).jpeg,4032,3024,tag,2103,1402,2499,2020
/content/NameTag (94).jpeg,4032,3024,tag,1485,1937,2055,2446
/content/NameTag (95).jpeg,4032,3024,tag,1785,1094,2094,1411
/content/NameTag (95).jpeg,4032,3024,tag,1799,1472,2312,2098
/content/NameTag (96).jpeg,4032,3024,tag,1042,2055,2177,2850
/content/NameTag (97).jpeg,4032,3024,tag,1272,311,1385,485
/content/NameTag (97).jpeg,4032,3024,tag,2581,602,2738,689
/content/NameTag (97).jpeg,4032,3024,tag,1881,2602,2012,2694
/content/NameTag (98).jpeg,4032,3024,tag,1259,2550,1485,2742
/content/NameTag (99).jpeg,4032,3024,tag,1807,389,2103,498

我正在 Google Colab 中运行我的模型制作器，/content/ 是我的图像所在的位置。

这是我的模型制作者代码

import numpy as np
import os

from tflite_model_maker.config import ExportFormat
from tflite_model_maker import model_spec
from tflite_model_maker import object_detector

import tensorflow as tf
assert tf.__version__.startswith('2')

tf.get_logger().setLevel('ERROR')
from absl import logging
logging.set_verbosity(logging.ERROR)

spec = model_spec.get('efficientdet_lite0')

train_data, validation_data, test_data = object_detector.DataLoader.from_csv('/content/test.csv')

model = object_detector.create(train_data, model_spec=spec, epochs=50, batch_size=8, train_whole_model=True, validation_data=validation_data)

这是我遇到的错误

        AttributeError                            Traceback (most recent call 
    last)
    <ipython-input-8-cd74bf318e21> in <module>()
          1 
    ----> 2 model = object_detector.create(train_data, model_spec=spec, 
    epochs=50, batch_size=8, train_whole_model=True, 
    validation_data=validation_data)

    /usr/local/lib/python3.7/dist- packages/tensorflow_examples/lite/model_maker/core/task/object_detector.py in create(cls, train_data, model_spec, validation_data, epochs, batch_size, train_whole_model, do_train)
        281           model_spec.compat_tf_versions, compat.get_tf_behavior()))
        282 
    --> 283     object_detector = cls(model_spec, train_data.label_map, 
    train_data)
        284 
        285     if do_train:

    AttributeError: 'NoneType' object has no attribute 'label_map'`

Answer 1

您的 CSV 格式似乎不正确。

CSV 的结构应如下所示：

TRAINING,a.jpg,NIK,0.03248433451607413,0.03522800346688288,,,0.08648954064904738,0.05116448122571085,,
TRAINING,a.jpg,Name,0.03309341578825052,0.05731540246596025,,,0.060705100126913526,0.06961724494645903,,
TRAINING,b.jpg,NIK,0.03454945054945055,0.03879072581362658,,,0.08973883259597545,0.06085517535899215,,
TRAINING,b.jpg,Name,0.03757813614956471,0.06583747041762307,,,0.05642329099471957,0.0807843555935159,,
TEST,sam_45.jpg,Name,0.0209833413323341,0.03232098987112413,,,0.04926242903560284,0.03803429616147435,,
TEST,sam_46.jpg,NIK,0.019437571862680685,0.02024184021555901,,,0.053739169267411305,0.02787400947716322,,
VALIDATION,sam_47.jpg,NIK,0.03698836813160072,0.013414367060884802,,,0.09344429843772814,0.01615199299167762,,
VALIDATION,sam_47.jpg,NIK,0.03966515793059814,0.06077529566360052,,,0.09466102107363604,0.06378668418747263,,

在第一列中，您应该定义

TRAINING

、

TEST

或

VALIDATION

。

和

的值应与图像大小相关，范围在0和1之间。在

x_min, y_min

和

x_max, y_max

之间以及末尾添加两个空逗号。格式应为

x_min, y_min,,,x_max, y_max,,

而不是

xmin, ymin, xmax, ymax

。另外，请记住从 CSV 中删除标头。

Tensorflow Lite 对象检测 csv

问题描述投票：0回答：1

1个回答

最新问题

Tensorflow Lite 对象检测 csv

问题描述 投票：0回答：1

1个回答

最新问题

问题描述投票：0回答：1