Hello,
I would like to detect bib (number) for a running race, with OpenCV.
First, in this image I would like to be able to find :
1192, 1177, 277, 290...
Then I would like to film runners, and to be able to all the runners at one race point. The camera won't be at the departure, so there will be less density of runners.
I downloaded OpenCV for .NET (Emgu) and I found in samples 2 approaches to do that :
1 - Plate recognition : you try to find to number (white rectangle, as the plate), and then you use OCR for getting the number. This option seems to be slow with the OCR.
2 - TrafficSignRecognition : I remplace sign by the digits (0-1-...9) as images, then I try to find them in the initial image
What do you think of these 2 approaches ? Do you think OpenCV meets my needs ? I would be able to detect numbers quite fast (like a treatment in 100ms) Do you know if there is similar project as mine ? I'm trying to do a prototype, but I start from nothing in Image Analysis so I would not complain to have some help :-)
Thanks in advance. Laurent (from France)