I'd go with that definition but it is more complex in reality.
The effective pixel count is that after sensor interpolation and processing from the Bayer pattern (or similar for Fovean etc) that transforms physical sensor locations into pixels in an RGB image. There isn't always a 1:1 correlation - it may be more or less. For example, Nikon D1X physical sensor is 5.33 million sensor locations that are interpolated in the Y axis and processed from a greater number of sensor locations in the X axis resulting in 5.89 million pixels (effective pixel count). Similarly D2H has 4.26 photo detectors for 4.0 million pixels, Canon 1DII has 8.5m for 8.2m pixels, etc.
For a better description try: Effective pixel description