Что может являться причиной остановки обучений opencv_haartraining?

Question

[ivandzemianchyk]

Функция запуска

opencv_haartraining -data data/1.1v4/ -vec pos/vec1.1.vec -npos 1000 -bg neg.txt -nneg 1466 -nstages 20 -w 30 -h 30

Вывод консоли

Data dir name: data/1.1v4/
Vec file name: pos/vec1.1.vec
BG  file name: neg.txt, is a vecfile: no
Num pos: 1000
Num neg: 1466
Num stages: 20
Num splits: 1 (stump as weak classifier)
Mem: 200 MB
Symmetric: TRUE
Min hit rate: 0.995000
Max false alarm rate: 0.500000
Weight trimming: 0.900000
Equal weights: FALSE
Mode: BASIC
Width: 30
Height: 30
Applied boosting algorithm: GAB
Error (valid only for Discrete and Real AdaBoost): misclass
Max number of splits in tree cascade: 0
Min number of positive samples per cluster: 500
Required leaf false alarm rate: 9.53674e-07

Tree Classifier
Stage
+---+
|  0|
+---+


Number of features used : 206475

Parent node: NULL

*** 1 cluster ***
POS: 1000 1000 1.000000
NEG: 1466 1
BACKGROUND PROCESSING TIME: 0.00
Precalculation time: 0.00
+----+----+-+---------+---------+---------+---------+
|  N |%SMP|F|  ST.THR |    HR   |    FA   | EXP. ERR|
+----+----+-+---------+---------+---------+---------+
|   1|100%|-|-0.974735| 1.000000| 1.000000| 0.041363|
+----+----+-+---------+---------+---------+---------+
|   2|100%|+|-0.117515| 0.999000| 0.104366| 0.041363|
+----+----+-+---------+---------+---------+---------+
Stage training time: 247.00
Number of used features: 2

Parent node: NULL
Chosen number of splits: 0

Total number of splits: 0

Tree Classifier
Stage
+---+
|  0|
+---+

   0


Parent node: 0

*** 1 cluster ***
POS: 1000 1001 0.999001
NEG: 1466 0.0759074
BACKGROUND PROCESSING TIME: 0.00
Precalculation time: 0.00
+----+----+-+---------+---------+---------+---------+
|  N |%SMP|F|  ST.THR |    HR   |    FA   | EXP. ERR|
+----+----+-+---------+---------+---------+---------+
|   1|100%|-|-0.874709| 1.000000| 1.000000| 0.085969|
+----+----+-+---------+---------+---------+---------+
|   2|100%|+|-0.173274| 0.997000| 0.257844| 0.063666|
+----+----+-+---------+---------+---------+---------+
Stage training time: 242.00
Number of used features: 2

Parent node: 0
Chosen number of splits: 0

Total number of splits: 0

Tree Classifier
Stage
+---+---+
|  0|  1|
+---+---+

   0---1


Parent node: 1

*** 1 cluster ***
POS: 1000 1004 0.996016
NEG: 1466 0.0205997
BACKGROUND PROCESSING TIME: 1.00
Precalculation time: 0.00
+----+----+-+---------+---------+---------+---------+
|  N |%SMP|F|  ST.THR |    HR   |    FA   | EXP. ERR|
+----+----+-+---------+---------+---------+---------+
|   1|100%|-|-0.947455| 1.000000| 1.000000| 0.107867|
+----+----+-+---------+---------+---------+---------+
|   2|100%|+|-0.451255| 0.999000| 0.389495| 0.062855|
+----+----+-+---------+---------+---------+---------+
Stage training time: 238.00
Number of used features: 2

Parent node: 1
Chosen number of splits: 0

Total number of splits: 0

Tree Classifier
Stage
+---+---+---+
|  0|  1|  2|
+---+---+---+

   0---1---2


Parent node: 2

*** 1 cluster ***
POS: 1000 1005 0.995025
NEG: 1466 0.0122045
BACKGROUND PROCESSING TIME: 2.00
Precalculation time: 0.00
+----+----+-+---------+---------+---------+---------+
|  N |%SMP|F|  ST.THR |    HR   |    FA   | EXP. ERR|
+----+----+-+---------+---------+---------+---------+
|   1|100%|-|-0.944708| 1.000000| 1.000000| 0.061233|
+----+----+-+---------+---------+---------+---------+
|   2|100%|+|-0.250223| 1.000000| 0.583220| 0.036902|
+----+----+-+---------+---------+---------+---------+
|   3| 75%|-|-1.009707| 0.998000| 0.113233| 0.036496|
+----+----+-+---------+---------+---------+---------+
Stage training time: 257.00
Number of used features: 3

Parent node: 2
Chosen number of splits: 0

Total number of splits: 0

Tree Classifier
Stage
+---+---+---+---+
|  0|  1|  2|  3|
+---+---+---+---+

   0---1---2---3


Parent node: 3

*** 1 cluster ***
POS: 1000 1007 0.993049
NEG: 1466 0.00156556
BACKGROUND PROCESSING TIME: 17.00
Precalculation time: 0.00
+----+----+-+---------+---------+---------+---------+
|  N |%SMP|F|  ST.THR |    HR   |    FA   | EXP. ERR|
+----+----+-+---------+---------+---------+---------+
|   1|100%|-|-0.737728| 1.000000| 1.000000| 0.113950|
+----+----+-+---------+---------+---------+---------+
|   2|100%|+|-1.678201| 1.000000| 1.000000| 0.076237|
+----+----+-+---------+---------+---------+---------+
|   3|100%|-|-0.993676| 1.000000| 0.453615| 0.064071|
+----+----+-+---------+---------+---------+---------+
Stage training time: 359.00
Number of used features: 3

Parent node: 3
Chosen number of splits: 0

Total number of splits: 0

Tree Classifier
Stage
+---+---+---+---+---+
|  0|  1|  2|  3|  4|
+---+---+---+---+---+

   0---1---2---3---4


Parent node: 4

*** 1 cluster ***
POS: 1000 1007 0.993049
NEG: 1466 0.00069932
BACKGROUND PROCESSING TIME: 38.00
Precalculation time: 0.00
+----+----+-+---------+---------+---------+---------+
|  N |%SMP|F|  ST.THR |    HR   |    FA   | EXP. ERR|
+----+----+-+---------+---------+---------+---------+
|   1|100%|-|-0.666781| 1.000000| 1.000000| 0.132198|
+----+----+-+---------+---------+---------+---------+
|   2|100%|+|-1.438892| 1.000000| 1.000000| 0.132198|
+----+----+-+---------+---------+---------+---------+
|   3|100%|-|-0.781253| 0.996000| 0.477490| 0.059611|
+----+----+-+---------+---------+---------+---------+
Stage training time: 363.00
Number of used features: 3

Parent node: 4
Chosen number of splits: 0

Total number of splits: 0

Tree Classifier
Stage
+---+---+---+---+---+---+
|  0|  1|  2|  3|  4|  5|
+---+---+---+---+---+---+

   0---1---2---3---4---5


Parent node: 5

*** 1 cluster ***
POS: 1000 1011 0.989120
NEG: 1466 0.000360453
BACKGROUND PROCESSING TIME: 73.00
Precalculation time: 0.00
+----+----+-+---------+---------+---------+---------+
|  N |%SMP|F|  ST.THR |    HR   |    FA   | EXP. ERR|
+----+----+-+---------+---------+---------+---------+
|   1|100%|-|-0.673204| 1.000000| 1.000000| 0.151257|
+----+----+-+---------+---------+---------+---------+
|   2|100%|+|-1.243254| 1.000000| 1.000000| 0.155718|
+----+----+-+---------+---------+---------+---------+
|   3|100%|-|-0.616460| 0.998000| 0.392906| 0.070154|
+----+----+-+---------+---------+---------+---------+
Stage training time: 294.00
Number of used features: 3

Parent node: 5
Chosen number of splits: 0

Total number of splits: 0

Tree Classifier
Stage
+---+---+---+---+---+---+---+
|  0|  1|  2|  3|  4|  5|  6|
+---+---+---+---+---+---+---+

   0---1---2---3---4---5---6


Parent node: 6

*** 1 cluster ***
POS: 1000 1013 0.987167

Answer 1

[ivandzemianchyk]

Да, действительно забыл добавить. Позитивные образцы были выгенерированы автоматический с использованием параметров

-maxidev 60 -maxxangle 0.2 -maxyangle 0.3 -maxzangle 0.2 -w 40 -h 40 -bgcolor 100 -bgthresh 20

Образцов было 1620.

Comments

[ivandzemianchyk]

Да, я читал статью. Но мне кажется что проблема в чём-то другом и связана скорее с параметрами:

Min hit rate: 0.995000
Max false alarm rate: 0.500000
Weight trimming: 0.900000

Как писал выше, я использовал 1600 образцов, а в параметрах указал 1000

[ivandzemianchyk]

1620 - это кол-во сгенерированных сэмплов из одного образца, генерировал с помощью createsamples

[ivandzemianchyk]

спасибо за помощь. По сути разные -w и -h не имеют значения т.к. и так происходит resizing