12-19 高斯灰度化修正检测数据记录

学术 实验
学术
发布日期:   2022-12-19
文章字数:   3.7k
阅读时长:   21 分
阅读次数:  

高斯灰度化修正检测数据记录

Kitti数据

0020

低可见度低阈值 高可见度高阈值

阈值 MIOU Precision recall
<= 0 0.8203 0.5988 0.5120
<= 0.2
<= 0.35
<= 0.4 0.8203 0.5988 0.5120
<= 0.42
<= 0.45 0.8203 0.5988 0.5120
<= 0.6 0.8202 0.5982 0.5069
<= 0.7 0.8227 0.5315 0.4476
<= 0.8 0.8241 0.6069 0.4368
<= aver(这组数据的平均灰度值 0.8459468438) 0.8264 0.7007 0.4130

灰度化对比 （阈值选择 <= aver ）

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.8316 0.6295 0.5414 4.2820
F 0.8203 0.5988 0.5120 4.9825
AVOD 0.8295 0.7939 0.3957 4.1687
ShapeMean 0.8381 0.6584 0.5084 4.4245
Best parameters 0.8320 0.6400 0.5335 4.6407
ordFusion 0.8258 0.6417 0.5352 4.6029
all 0.8329 0.6422 0.5360 4.6407 0.8383 0.6688 0.4525
fus 0.8400 0.8794 0.4067 4.3230 0.8449 0.7957 0.3584
all_second 0.8273 0.6106 0.4176 4.6029 0.8330 0.6282 0.3474 4.4364
fus_second 0.8289 0.7105 0.4053 4.5059 0.8347 0.6926 0.3342 4.3230

image-20230105160937396

0018

低可见度低阈值 高可见度高阈值

阈值 MIOU Precision recall
<= 0 0.8493 0.6897 0.5883
<= 0.2
<= 0.35
<= 0.4 0.8493 0.6897 0.5883
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8 0.8494 0.7143 0.5876
<= aver(这组数据的平均灰度值 0.8797849462365619) 0.8358 0.0453 0.0402

灰度化对比 （阈值选择 <= 0.8）

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.8372 0.5144 0.5742 5.2677
F 0.8493 0.6897 0.5883 4.5802
AVOD 0.8669 0.7539 0.3864 4.7750
ShapeMean 0.8589 0.7001 0.6130 4.1864
Best parameters 0.8604 0.6919 0.6053 4.2792
ordFusion 0.8606 0.6955 0.6090 4.2792
all 0.8614 0.6920 0.6058 4.2792 0.8615 0.6917 0.6058
fus 0.8709 0.7831 0.5228 4.2640 0.8711 0.7831 0.5219
all_second 0.8671 0.8000 0.5625 4.1655 0.8664 0.7956 0.5640 4.3241
fus_second 0.8683 0.8673 0.5537 4.0967 0.8683 0.8695 0.5522 4.2640

0014

低可见度低阈值 高可见度高阈值

阈值 MIOU Precision recall
<= 0 0.7983 0.2750 0.2290
<= 0.2
<= 0.35
<= 0.4 0.7990 0.2731 0.2271
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8 0.8010 0.2953 0.2123
<= aver(这组数据的平均灰度值 0.8197169811320755) 0.8010 0.3026 0.2123

灰度化对比 （阈值选择 <= 0.8197169811320755）

无AVOD对比算法数据

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.7626 0.1091 0.1043 6.7845
F 0.7900 0.2750 0.2290 5.5987
AVOD 0.7793 0.1863 0.1254 6.3586
ShapeMean 0.7925 0.2698 0.2201 5.8267
Best parameters 0.7972 0.2882 0.2336 5.6352
ordFusion 0.7982 0.2836 0.2300 5.9610
all 0.7977 0.2852 0.2312 5.6352 0.7985 0.2824 0.2169
fus 0.7981 0.1944 0.1079 5.4905 0.7998 0.1703 0.0935
all_second 0.7845 0.1868 0.1752 5.9610 0.7897 0.1585 0.1215 5.2702
fus_second 0.7888 0.2129 0.1615 6.1237 0.7981 0.1944 0.1079 5.4905

0011

低可见度低阈值 高可见度高阈值

阈值 MIOU Precision recall
<= 0 0.8087 0.6011 0.4715
<= 0.2
<= 0.35
<= 0.4
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8 0.8097 0.5978 0.4482
<=0.9 0.8126 0.3994 0.2755
<= aver(这组数据的平均灰度值 0.9289008042895465) 0.8109 0.2742 0.1972

灰度化对比 （阈值选择 <= 0.8）

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.8328 0.6044 0.5202 5.2677
F 0.8087 0.6011 0.4715 5.3947
AVOD 0.8260 0.6485 0.3620 4.8097
ShapeMean 0.8274 0.5975 0.5091 4.8483
Best parameters 0.8196 0.5916 0.5059 4.7750
ordFusion 0.8215 0.5916 0.5059 5.1182
all 0.8219 0.5925 0.5066 5.1182 0.8226 0.5632 0.4854
fus 0.8332 0.8181 0.4182 4.6838 0.8340 0.7909 0.3961
all_second 0.8247 0.5589 0.3378 5.5732 0.8248 0.5353 0.3254 5.4942
fus_second 0.8268 0.6955 0.3230 5.6386 0.8273 0.6765 0.3106 5.5078

0009

低可见度低阈值 高可见度高阈值

阈值 MIOU Precision recall
<= 0 0.7986 0.5385 0.4592
<= 0.2
<= 0.35
<= 0.4 0.7998 0.4408 0.3671
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8 0.8063 0.2052 0.1594
<=0.9
<= aver(这组数据的平均灰度值 0.7146743295019157) 0.8008 0.2836 0.2256

灰度化对比 （阈值选择 <= 0.4）

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.8023 0.5695 0.4716 5.4125
F 0.7986 0.5385 0.4592 5.8082
AVOD 0.7919 0.6253 0.2900 5.9729
ShapeMean 0.7995 0.4693 0.3781 5.8934
Best parameters 0.8017 0.5646 0.3680 5.9873
ordFusion 0.8018 0.5662 0.3694 5.9341
all 0.8016 0.5642 0.4840 5.7364 0.8055 0.1835 0.1646
fus 0.8083 0.7207 0.3677 5.6055 0.8173 0.2600 0.1147
all_second 0.7963 0.4246 0.2644 5.9921 0.8060 0.1507 0.1019 5.7100
fus_second 0.7977 0.4989 0.2530 5.9257 0.8068 0.1884 0.0962 5.6055

0008

低可见度低阈值 高可见度高阈值

阈值 MIOU Precision recall
<= 0 0.7995 0.2963 0.2994
<= 0.2
<= 0.35
<= 0.4 0.7992 0.2642 0.2722
<= 0.42
<= 0.45 0.7992 0.2655 0.2722
<= 0.5 0.8010 0.2498 0.2608
<= 0.6 0.8012 0.2547 0.2582
<= 0.7 0.8034 0.2191 0.2231
<= 0.8 0.8037 0.2402 0.2183
<=0.9
**<= aver(这组数据的平均灰度值 0.792487179487184) 0.8037 0.2360 0.2183

灰度化对比 （阈值选择 <= 0.6） fp 取 336

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.8128 0.2977 0.2480 3.6941
F 0.8004 0.2769 0.2927 4.6500
AVOD 0.8106 0.3264 0.1889 3.7520
ShapeMean 0.8058 0.3210 0.2783 4.1365
Best parameters 0.8040 0.2910 0.2528 4.3122
ordFusion 0.8036 0.2938 0.2551 4.3122
all 0.8040 0.2938 0.2551 4.3122 0.8058 0.2938 0.2551
fus 0.8137 0.4252 0.1670 3.7988 0.8156 0.4252 0.1841
all_second 0.7936 0.2227 0.2222 4.5574 0.8128 0.3010 0.1874 4.1813
fus_second 0.7943 0.2894 0.2188 4.4268 0.8137 0.4252 0.1841 3.7988

0004

低可见度低阈值 高可见度高阈值

阈值 MIOU Precision recall
<= 0 0.7929 0.3477 0.3864
<= 0.2
<= 0.35
<= 0.4 0.7929 0.3477 0.3864
<= 0.42
<= 0.45
<= 0.5
<= 0.6 0.7925 0.3333 0.3660
<= 0.7
<= 0.8 0.7903 0.2390 0.2673
<=0.9
**<= aver(这组数据的平均灰度值 0.8174840764331214) 0.7903 0.2417 0.2673

灰度化对比 （阈值选择 <= 0.6） fp 取 286

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.8066 0.1105 0.2483 4.4080
F 0.7900 0.3064 0.3524 4.2825
AVOD 0.7919 0.6253 0.2900 3.9407
ShapeMean 0.7950 0.1898 0.2221 4.9016
BestParameters 0.7926 0.1879 0.2188 4.4894
ordFusion 0.7934 0.1872 0.2179 4.4894
all 0.7941 0.1883 0.2189 4.4894 0.7786 0.1039 0.1220
fus 0.7973 0.1088 0.0775 5.0708 0.7833 0.0719 0.0561
all_second 0.7906 0.1159 0.1586 4.5748 0.7912 0.0760 0.0811 5.5909
fus_second 0.7923 0.1882 0.1536 4.3392 0.7950 0.1056 0.0761 5.0708

效果差

自采数据

高斯化灰度值 等级分布

grayDate = {1: 0.02, 2: 0.11, 3: 0.42, 4: 0.68, 5: 0.8, 6: 0.91, 7: 0.99, 8: 0.99, 9: 0.87, 10: 0.14}

104

阈值 MIOU Precision recall
<= 0 0.8037 0.5235 0.3180
<= 0.2 0.8037 0.5235 0.3180
<= 0.35
<= 0.4 0.8043 0.5334 0.3167
<= 0.42 0.8029 0.3534 0.2072
<= 0.45 0.8029 0.3534 0.2072
<= 0.6
<= 0.7
<= 0.8
<= aver(这组数据的平均灰度值 0.6845882352941179)

灰度化对比 （阈值选择 <= 0.4） fp 取83 灰度化融合时取 73

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.7862 0.2379 0.3002 4.9977
F 0.8040 0.5241 0.3156 4.0917
AVOD 0.7673 0.2683 0.0879 6.0681
ShapeMean 0.8011 0.4608 0.3184 4.0134
Best parameters 0.8105 0.4619 0.3720 3.8893
ordFusion
all 0.8119 0.4526 0.3625 3.8893 0.8266 0.1561 0.1047 4.0308
fus 0.8179 0.5964 0.3067 3.7985 0.8305 0.1878 0.0946 3.8513
all_second 0.8105 0.3865 0.1258 5.2508 0.8291 0.0926 0.0477 3.9820
fus_second 0.7719 0.5964 0.0326 6.0042 0.7865 0.0579 0.0101 3.7985

![](E:\魏泽太学术\05毕业论文\03 大论文\visio图\WZT00104.jpg)

2023.2.25 数据

105

阈值 MIOU Precision recall
<= 0 0.7940 0.4303 0.3132
<= 0.2 0.8020 0.1893 0.1722
<= 0.35
<= 0.4 0.8010 0.2017 0.1695
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8
<= aver(这组数据的平均灰度值 0.27) 0.8020 0.1893 0.1722

灰度化对比 （阈值选择 <= 0.2） fp = 120 IOU = 0.5

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.7983 0.1850 0.3397 5.0717
F 0.7939 0.4305 0.3058 4.5551
AVOD 0.7828 0.1336 0.0727 7.1912
ShapeMean
Best parameters
ordFusion
all 0.8135 0.3681 0.3255 4.3209 0.8193 0.3681 0.3255
fus 0.8172 0.4874 0.2824 4.2416 0.8238 0.4874 0.2824
all_second 0.7796 0.2124 0.2313 4.9205 0.8160 0.3186 0.3192 4.4582
fus_second 0.7691 0.2892 0.1862 4.8999 0.8172 0.4874 0.2824 4.2416

image-20230225215958176

![](E:\魏泽太学术\05毕业论文\03 大论文\visio图\WZT00105.jpg)

106

阈值 MIOU Precision recall
<= 0 0.7629 0.0232 0.0078
<= 0.2
<= 0.35
<= 0.4 0.7629 0.0232 0.0078
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8
<= aver(这组数据的平均灰度值 0.594090909090909) 0.7603 0.0210 0.0066

灰度化对比 （阈值选择 <= 0.4）

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.7835 0.0062 0.0042 6.2077
F 0.7629 0.0232 0.0078 5.4167
AVOD 0.7389 0.0282 0.0033 9.8645
ShapeMean
Best parameters
ordFusion
all 0.7824 0.0118 0.0054 7.0656 0.7771 0.0163 0.0066
fus 0.7824 0.0167 0.0054 7.0656 0.7824 0.0159 0.0054
all_second 0.8047 0.0132 0.0065 6.6001 0.8160 0.0108 0.0053 7.4849
fus_second 0.8503 0.0115 0.0021 5.6537 0.8503 0.0128 0.0021 7.0656

image-20230225220136020

107

阈值 MIOU Precision recall
<= 0 0.7994 0.3567 0.1985
<= 0.2
<= 0.35
<= 0.4 0.8184 0.1972 0.0629
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8
<= aver(这组数据的平均灰度值 0.23674242424242437) 0.8157 0.1710 0.0663

灰度化对比 （阈值选择 <= 0.4 ）fp = 107

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.7783 0.2301 0.2880 5.5163
F 0.7926 0.3465 0.1965 4.1564
AVOD 0.8004 0.1464 0.0594 6.6100
ShapeMean
Best parameters
ordFusion
all 0.7978 0.3123 0.2250 4.1892 0.8130 0.1250 0.0593
fus 0.7972 0.3022 0.1308 4.0599 0.8179 0.1546 0.0345
all_second 0.7802 0.1122 0.1171 5.8315 0.7955 0.2451 0.1736 4.4528
fus_second 0.7876 0.1195 0.0852 6.4538 0.7972 0.3022 0.1308 4.0599

image-20230225220255785

109

阈值 MIOU Precision recall
<= 0 0.7762 0.0629 0.0226
<= 0.2
<= 0.35
<= 0.4 0.8245 0.0036 0.0009
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8
<= aver(这组数据的平均灰度值 0.07187943262411328) 0.7864 0.0249 0.0101

灰度化对比 （阈值选择 <= 0 ） fp = 92

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.7926 0.2833 0.3861 5.2864
F 0.7338 0.0110 0.0027 7.6199
AVOD 0.7839 0.0711 0.0182 6.9643
ShapeMean 0.7934 0.5306 0.3361 5.2786（7.8121）
Best parameters 0.8066 0.5306 0.3361 5.2970
ordFusion
all 0.8188 0.5306 0.3361 5.2970 0.7993 0.0865 0.0633
fus 0.8079 0.0330 0.0077 8.4705 0.8079 0.0330 0.0077
all_second 0.7686 0.2466 0.2209 6.0361 0.7714 0.1102 0.0553 5.5375
fus_second 0.7657 0.2913 0.1733 6.4776 0.8079 0.0330 0.0077 8.4705

image-20230225220505199

灰度化对比 （阈值选择 <= 0 ） fp = 260

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.7995 0.2062 0.3393 5.3160
F 0.7767 0.0644 0.0232 6.0716
AVOD 0.8106 0.0848 0.0276 6.9131
ShapeMean 0.8041 0.7812 0.3947 0.1100 0.2758 0.0385 5.1623 5.3693
Best parameters 0.8062 0.7975 0.3896 0.0965 0.2727 0.0354 5.1754 5.4023
ordFusion
all 0.8060 0.3878 0.2687 5.1754 0.7993 0.0865 0.0633
fus 0.7948 0.0894 0.0314 5.4023 0.8079 0.0330 0.0077
all_second 0.7887 0.1765 0.1657 5.9212 0.8008 0.1161 0.0838 5.5276
fus_second 0.7782 0.2157 0.1133 6.0677 0.7948 0.0894 0.0314 5.4023

110

阈值 MIOU Precision recall
<= 0 0.7646 0.0906 0.0284
<= 0.2 NAN
<= 0.35
<= 0.4
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8
<= aver(这组数据的平均灰度值 0.02346153846153847) NAN

灰度化对比 （阈值选择 <= 0 ） fp = 63

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.7851 0.1699 0.2517 5.3698
F 0.7662 0.1344 0.0452 3.2116
AVOD 0.7545 0.1559 0.0485 5.7982
ShapeMean
Best parameters
ordFusion
all 0.7791 0.1239 0.0718 4.0371 0.7791 0.1212 0.0718
fus 0.7722 0.1048 0.0288 4.0877 0.7722 0.1048 0.0288
all_second 0.8132 0.0347 0.0296 4.4623 0.7946 0.0379 0.0328 4.2909
fus_second 0.7581 0.0161 0.0065 5.1175 0.7298 0.0242 0.0097 4.0877

112

阈值 MIOU Precision recall
<= 0 0.7897 0.0784 0.0343
<= 0.2
<= 0.35
<= 0.4
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8
<= aver(这组数据的平均灰度值 0.024909090909090922) 0.8641 0.0196 0.0016

灰度化对比 （阈值选择 <= ）

oring MIOU Precision recall gray MIOU Precision recall MIOU Precision recall
P
F
AVOD
ShapeMean
Best parameters
ordFusion
all
fus
all_second
fus_second

image-20230225220920789

113

阈值 MIOU Precision recall
<= 0 0.7613 0.0223 0.0064
<= 0.2
<= 0.35
<= 0.4
<= 0.42
<= 0.45
<= 0.6
<= 0.7
<= 0.8
<= aver(这组数据的平均灰度值 0.020000000000000014) NAN

灰度化对比 （阈值选择 <= 0）fp = 99

oring MIOU Precision recall MCLE gray MIOU Precision recall MCLE MIOU Precision recall
P 0.7816 0.0367 0.0773 6.3496
F 0.7613 0.0255 0.0074 6.2258
AVOD 0.8422 0.0153 0.0083 6.2959
ShapeMean
Best parameters
ordFusion
all 0.7407 0.0187 0.0130 5.4643 0.7758 0.0323 0.0232
fus 0.7141 0.0187 0.0062 5.0000 0.7141 0.0187 0.0062
all_second 0.7483 0.0015 0.0017 6.5417 0.7875 0.0083 0.0058 5.4837
fus_second 0.7483 0.0034 0.0017 5.8696 0.7875 0.0102 0.0058 5.0000

image-20230225221030236

代码执行步骤

1》在 S00parameter.py 文件内设定参数

2》执行S0_main.py 文件 完成一阶段融合

3》执行S10_main.py 文件 完成一阶段+灰度化

4》执行S7_SecondFusion.py 文件 完成二阶段多帧融合

5》执行S10_SecondFusion_gray.py 文件 完成 二阶段+灰度化

其他代码说明：

S8_grayLevel.py 文件 进行灰度化等级划分

S9_grayLevelThreshold.py 文件 进行灰度化阈值设定

S10_main_gray_dis.py 文件 在雷达信息中附带距离信息，距离作用暂未完成

学术 实验
赏
评论
  目录