We are a game company that is producing arcade games for the likes of Dave and Busters. We currently have a "Ring Toss" game working with 80% 90% recognition rate (using openCV, but need to improve this to 95% to 99%. We do not have the in house expertise to take this to the next level and are looking for someone who can.
I have uploaded additional images:
We have a calibration routine that determines the center of the bottles and generate a text file with the data. (bottles.txt)
feOut++. jpg show bottle centers plotted with the ring centers. As shown, if the ring center is coincident+/- with the bottle center, it has a ring on it. This is the current method we are using, but does not work well if you have two rings overlapping in the same images in a series. We are using a background subtraction method.
Also include are two series of photos from two different game play, one where individual rings are being tossed and the other where groups of rings are bing tossed. These series are taken at .5 second intervals. The process must be quick enough on a Raspberry Pi 3 to evaluate each frame within this time constraint.
The project will be to take these series and evaluate them and at each frame time indicate which bottle has been "ringed" and how many rings it has.
Hours to be determined
1 to 3 months1-3 monthsProject LengthDuration
I am willing to pay higher rates for the most experienced freelancers
One-time Project:Project: IT & Software - Desktop Programming - Other