This paper designs and implements a cervical spondylosis prevention system based on pressure sensors and the computer camera. This system obtains the body pressure data through a cushion which contains many pressure sensors, and the human face data through a computer camera. We optimize the number and layout of pressure sensors in the cushion, and reduce the computational complexity by image processing by recognizing facial features. By fusing the data from both pressure sensors and camera, this system achieves a more accurate sitting posture recognition, comparting with using pressure sensors and camera separately. The results are further used to estimate the risk level of cervical spondylosis for the users. The system can also guide the users to complete a series of given actions to test their criticality or potential risk of cervical spondylosis. This system is smart, well-designed, and non-invasive, and hence quite applicable to common workplaces.