i am looking for some tool/software that can train HOG descriptor for some object. or if there is any straight tutorial for doing this, please provide the link of it. or suggest how can i train hog descriptor. Basically my aim is to detect the logo of the channel from the image of television.. the image of the program(going on television) is taken from camera. .. this image is the input to our program which is to identify the channel of which the image is taken. i tried surf/shift based feature but these are very slow for matching pattern with 3000 of images(10 images per channel).. whereas haar based feature take lot of time for training for more than 300 tv channels. We need near real time(< 2 sec) performance of program..
Asked
Active
Viewed 2,357 times
2 Answers
0
HOG for "detecting object" opencv
Read this one and my suggestion is please check at this forum for the same question,all ready many people answered the same question.
Anyway if you still did not understand the answers by any one, I will give you the whole code for training and testing.
-
Sir, it will be very helpful if u the give the whole code for training and testing(if possible with instructions). i am trying this for week but no avail. Thank you – user1441867 Jun 11 '12 at 06:08
-
1Sir, this is my email lovelylovvy@gmail.com , if you could email the code.. waiting for your reply. – user1441867 Jun 14 '12 at 09:20