Ask Your Question

Object detection in iOS using cascades

asked 2013-05-22 16:47:59 -0500

nes_4life gravatar image

I'm new to OpenCV but am a seasoned iOS programmer looking to create an iOS app that can detect when a specific type of object is present in a video feed and then go on to highlight that object - or at least give me the bounds of the object.

I've found various fragments of information on Cascade Classifier Training, Haar training, and openCV for iOS but found that nothing really covers the entire process. Could anyone point me in the right direction of a comprehensive tutorial of the procedure for making a cascade xml using a Mac and then the further implementation of object detection using that file in iOS? I am familiar with extracting frames from a vido feed in iOS, analysing images at a bit level, and also have a good grasp on the use of Haar wavelets for face recognition. My main downfall is that I have yet to find a clear and maintained example of openCV for iOS (that doesn't involve me attempting to compile openCV for iOS) and i'm confused on what my 1000-odd positive photos should actually contain (all cropped to same size, different lighting, angles, scale, other objects, background etc).

Many thanks for your time.

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted

answered 2013-06-19 01:19:23 -0500

simon111 gravatar image

Tranning is not very diffcult, but it spend long time (at least one week, depend on which computer you use), Opencv provides state-of-art tools for trainning. Check this link:

Samples is very important, you should prepare samples as many as possible. Maybe you need write a simple annotation tool to create samples.

edit flag offensive delete link more


A PR has been added for both 2.4 and master branch to get a universal annotation tool included in the repository.

StevenPuttemans gravatar imageStevenPuttemans ( 2015-01-29 06:57:50 -0500 )edit
Login/Signup to Answer

Question Tools

1 follower


Asked: 2013-05-22 16:47:59 -0500

Seen: 1,714 times

Last updated: Jun 19 '13