Ask Your Question

"Train dataset for temp stage can not be filled." on opencv_traincascade

asked 2015-07-31 20:23:48 -0600

e_ntyo gravatar image

I'm trying to make a Haar Feature-based Cascade Classifier using opencv_traincascade. But I have had this message.

PS C:\cascade2\TrainingData> opencv_traincascade -data .\result\ -vec positive.vec -bg .\negative.txt -numPos 4500 -numNeg 2355 -f HAAR -mode ALL


cascadeDirName: .\result\

vecFileName: positive.vec (positive.vec is made of 5000 JPEG files created by create_samples and one JPEG file.)

bgFileName: .\negative.txt

numPos: 4500

numNeg: 2355

numStages: 20

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

maxTreeDepth: 1

maxWeakCount: 100

mode: BASIC

Stages 0-9 are loaded

===== TRAINING 10-stage =====


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

This attempt is the second time. When I tried it for the first time, Anything goes from stage0 to stage9. But I received the same error on stage10. What should I do?

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted

answered 2016-06-10 08:49:13 -0600

cv_enthusiast gravatar image

Hey, i had the same problem like you!

First of all, I'm a bit confused... your input is

-mode ALL

but your output is

mode: BASIC

Anyway.. i fixed my problem by turning up both values

Kind regards

precalcValBufSize and precalcIdxBufSize

to 2048

I also just got this problem when i used -mode ALL

edit flag offensive delete link more

Question Tools

1 follower


Asked: 2015-07-31 20:16:09 -0600

Seen: 341 times

Last updated: Jun 10 '16