Error "Train dataset for temp stage can not be filled. Branch training terminated."

[romulogcerqueira]

I am trying to train a haar-like features using this tutorial, but I always have my training ended because this error:

"Train dataset for temp stage can not be filled. Branch training terminated."

How can I fix this?

[mrnugget]

Hey!

That's a pretty common error. See all these questions on the OpenCV board. There is not much I know about this error and I can only guess a solution, since there are a lot of possible reasons for that error.

And without any information about your setup there is nothing I can do.

[romulogcerqueira]

I have 55 positive images and 395 negative images (all images are 32x32).

I created my pos.dat and neg.dat files as:

$ find positive_images/ -name '.jpg' -exec echo {} 1 0 0 32 32 ; > pos.dat
$ find negative_images/ -name '.jpg' > neg.dat

Create my vec file as:
$ opencv_createsamples -info pos.dat -w 24 -h 24 -vec flowers.vec -num 55

And my training:
$ opencv_traincascade -data data_5 -vec flowers.vec -bg neg.dat -numStages 5 -numPos 55 -numNeg 395 -w 24 -h 24

And I got the following output error:
PARAMETERS:
cascadeDirName: data_5
vecFileName: flowers.vec
bgFileName: neg.txt
numPos: 55
numNeg: 395
numStages: 5
precalcValBufSize[Mb] : 256
precalcIdxBufSize[Mb] : 256
stageType: BOOST
featureType: HAAR
sampleWidth: 24
sampleHeight: 24
boostType: GAB
minHitRate: 0.995
maxFalseAlarmRate: 0.5
weightTrimRate: 0.95
maxDepth: 1
maxWeakCount: 100
mode: BASIC

===== TRAINING 0-stage =====
<BEGIN
POS count : consumed 55 : 55
NEG count : acceptanceRatio 395 : 1
Precalculation time: 2
+----+---------+---------+
| N | HR | FA |
+----+---------+---------+
| 1| 1| 1|
+----+---------+---------+
| 2| 1|0.0683544|
+----+---------+---------+
END>
Training until now has taken 0 days 0 hours 0 minutes 8 seconds.

===== TRAINING 1-stage =====
<BEGIN
POS count : consumed 55 : 55
NEG count : acceptanceRatio 395 : 0.123592
Precalculation time: 1
+----+---------+---------+
| N | HR | FA |
+----+---------+---------+
| 1| 1| 1|
+----+---------+---------+
| 2| 1| 0.265823|
+----+---------+---------+
END>
Training until now has taken 0 days 0 hours 0 minutes 16 seconds.

===== TRAINING 2-stage =====
<BEGIN
POS count : consumed 55 : 55
NEG count : acceptanceRatio 395 : 0.0283154
Required leaf false alarm rate achieved. Branch training terminated.

[mrnugget]

Sorry, I have no idea! Again, I can only guess.

And since this problem doesn't directly affect this project, I'm closing the issue for now.

[gloomyfish]

i have same problem, but when i put the bg.txt into build/bin directory, the error has gone, works fine now
i think it is a bug in opencv 3.1.0, since the -bg parameter did not support any path/dir, with the file name

[MahmoudAbdelRahman]

Hi,
I faced the same problem and was solved by modifying the bg.txt file . I've written the whole address to the bg images as follows :

E:\RProj\ALL_LABELS\trees\negative\img_0.jpg
E:\RProj\ALL_LABELS\trees\negative\img_1.jpg
E:\RProj\ALL_LABELS\trees\negative\img_2.jpg
E:\RProj\ALL_LABELS\trees\negative\img_3.jpg
E:\RProj\ALL_LABELS\trees\negative\img_4.jpg
E:\RProj\ALL_LABELS\trees\negative\img_5.jpg
E:\RProj\ALL_LABELS\trees\negative\img_6.jpg
E:\RProj\ALL_LABELS\trees\negative\img_7.jpg
E:\RProj\ALL_LABELS\trees\negative\img_8.jpg
E:\RProj\ALL_LABELS\trees\negative\img_9.jpg
E:\RProj\ALL_LABELS\trees\negative\img_10.jpg
E:\RProj\ALL_LABELS\trees\negative\img_11.jpg
E:\RProj\ALL_LABELS\trees\negative\img_12.jpg
E:\RProj\ALL_LABELS\trees\negative\img_13.jpg
E:\RProj\ALL_LABELS\trees\negative\img_14.jpg
.
.
.
.

[DamonDBT]

also have this problem and do not find a way to handle this.

[rbk1234567]

Hello,
I want to confirm that paths in the bg.txt file need to be absolute.
On linux I was forced to use paths as /home/..../image.jpg to don't get this error.
Adding path as -bg parameter didn't work.
Used OpenCV 3.4.10 build

[ignacio-escalera]

Hi,
I faced the same problem and was solved by modifying the bg.txt file . I've written the whole address to the bg images as follows :

E:\RProj\ALL_LABELS\trees\negative\img_0.jpg
E:\RProj\ALL_LABELS\trees\negative\img_1.jpg
E:\RProj\ALL_LABELS\trees\negative\img_2.jpg
E:\RProj\ALL_LABELS\trees\negative\img_3.jpg
E:\RProj\ALL_LABELS\trees\negative\img_4.jpg
E:\RProj\ALL_LABELS\trees\negative\img_5.jpg
E:\RProj\ALL_LABELS\trees\negative\img_6.jpg
E:\RProj\ALL_LABELS\trees\negative\img_7.jpg
E:\RProj\ALL_LABELS\trees\negative\img_8.jpg
E:\RProj\ALL_LABELS\trees\negative\img_9.jpg
E:\RProj\ALL_LABELS\trees\negative\img_10.jpg
E:\RProj\ALL_LABELS\trees\negative\img_11.jpg
E:\RProj\ALL_LABELS\trees\negative\img_12.jpg
E:\RProj\ALL_LABELS\trees\negative\img_13.jpg
E:\RProj\ALL_LABELS\trees\negative\img_14.jpg
.
.
.
.

I was looking through the same situation and thankfully this was the answer, thanks!