Made changes to allow ml module to work with big data. #395

Merged
merged 3 commits into from Feb 4, 2013

Conversation

6 participants
@LeonidBeynenson

No description provided.

@ghost ghost assigned vpisarev Jan 31, 2013

@vpisarev

This comment has been minimized.

Show comment Hide comment
@vpisarev

vpisarev Jan 31, 2013

Contributor

Leonid, could you please correct the warnings: http://pullrequest.opencv.org

Contributor

vpisarev commented Jan 31, 2013

Leonid, could you please correct the warnings: http://pullrequest.opencv.org

@@ -796,7 +796,7 @@ struct CV_EXPORTS CvDTreeTrainData
const CvMat* responses;
CvMat* responses_copy; // used in Boosting
- int buf_count, buf_size;
+ int buf_count, buf_size; // buf_size is obsolete, please do not use it, use expression ((int64)buf->rows * (int64)buf->cols / buf_count) instead

This comment has been minimized.

Show comment Hide comment
@taka-no-me

taka-no-me Jan 31, 2013

Contributor

Can this field be completely removed in master?

@taka-no-me

taka-no-me Jan 31, 2013

Contributor

Can this field be completely removed in master?

@vpisarev

This comment has been minimized.

Show comment Hide comment
@vpisarev

vpisarev Jan 31, 2013

Contributor

yes, removing buf_size is probably a good idea.

Contributor

vpisarev commented Jan 31, 2013

yes, removing buf_size is probably a good idea.

LeonidBeynenson added some commits Feb 1, 2013

@vpisarev

This comment has been minimized.

Show comment Hide comment
@vpisarev

vpisarev Feb 4, 2013

Contributor

👍

Contributor

vpisarev commented Feb 4, 2013

👍

@opencv-pushbot opencv-pushbot merged commit 87b0126 into opencv:2.4 Feb 4, 2013

1 check was pending

default precommit_docs was executed, precommit_android was executed, precommit_windows was executed, precommit_ubuntu was executed
Details
@samuel1208

This comment has been minimized.

Show comment Hide comment
@samuel1208

samuel1208 Apr 16, 2014

I use 20000 faces positive sample to train face detector. If I use decision tree with depth 2 as the weaker classifier, it will at least 100 weaker classifiers to achieve 0.9 TP and 0.3FP . Is that right. I don't know if it's OK to post question here.

I use 20000 faces positive sample to train face detector. If I use decision tree with depth 2 as the weaker classifier, it will at least 100 weaker classifiers to achieve 0.9 TP and 0.3FP . Is that right. I don't know if it's OK to post question here.

@Beenking

This comment has been minimized.

Show comment Hide comment
@Beenking

Beenking Dec 25, 2014

orz

orz

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment