The project is full time and would prefer here but could be done remotely.
The project is to enhance our existing (or replace) tracking application which identifies the area a separate camera should be capturing and streaming. this is used for sporting events like hockey, basketball, soccer, etc.
Examples can be found here:
The existing application which uses input from either a panamorphic or multi-sensor cameras (already stitched) uses opencv to identify areas of movement and based on a predefined zone setting will send REST commands to an external device (usually a ptz camera to go to a preset).
It has worked pretty well, but we are looking to make the camera motion a bit nicer by enhancing the tracking. By nicer I mean provide more input to the camera to make the motion smoother and to detect players, identify team colors and numbers).
Our real goal would be to use a single multi-sensor camera and create a virtual ptz from the same input video stream and stream that instead of using the separate camera.
We have dozens of locations each with one of 3 different capture cameras so the application needs to be configurable for each (a type of calibration routine). Each venue has a i7 16GB server with SSD, running a virtual environment in linux.
As far as qualifications go, I need someone how has extensive knowledge in opencv, I don’t care if it’s used in c++ or python but they must have already done work on object detection and tracking. Good communication skills and ideally understands sports like hockey, basketball and hockey so that they understand what should be the region of interest. As mentioned the environment is linux.
As for remote work, I don’t have a problem with as long as they keep lines of communication open, and asks lots of questions and they have good english skills.