Actual Video Analysis (difficult)
This feature involves video analysis, a leap forward in difficulty. Only makes any sense for SC if there is high demand or you can charge much more.
I would like a feature that monitors the position of the head, or maybe a leg or maybe a hip during a swing.
The point of the feature is to provide instant feedback for a violation of physical motion limits drawn on the screen. In golf this might be lateral torso sway or lateral head movement. In this way the video analysis occurring in real time in the 'live' window becomes a training aid, not just a post-shot-review aid.
I see it implemented as follows
1 Define the area to be monitored using the drawing tools on a live screen. Let’s say golfers head with a circle sized in diameter to the head.
2. Define a no-fly/Red Zone using drawing tools behind or ahead of the head or even above the head for early extension problems. Red boxes perhaps.
Then….
A. If the head (through pattern recognition software) moves into the red zone on a practice swing (no ball nor trigger needed), a ‘warning’ sound is played.
OR
B. If the head (..see A above..) moves at all then an increasing frequency tone is played as the distance between the ‘object(head)’ and the Red Zone decrease.
This is real time feedback for training on practice swings.
With image analysis software (not unlike done with surveillance camera where the background is stationary and one object is moving) Swing Catalyst could substitute for a number of golf training devices (and charge more money of course for RedZone).
A simple start for this feature…and to build out your software library, would be to auto-detect the three swing points used as references. A white ball would have to be in view, initially you might require a reflective tape strip on the club itself in development, and work backward from impact (or forward from takeaway) to find the swing transition point. You could likely calculate the angle of the club shaft at the top, as well, detecting over-parallel for example. Reflective stripe on hat and glove might help, or reflective belt for torso angle/rotation.
You would have to require the camera with suitable resolution camera where each pixel represents a suitable physical distance. Because this is a significant software investment and a major upgrade, I would choose a camera you like and restrict to that. the software problems are big enough without camera differences. The software upgrade includes the cameras to force the issue for the first few years? Eventually, many of the angles and movements that are teased out with drawing tools can be automated.
I see this is as a several year development project in a strategic plan.