This work presents a computer vision image-based methodology for quantitative wear particle diagnostics in polymer composite gear systems. Conventional monitoring techniques typically provide limited information on the morphology and size distribution of the particles, restricting their diagnostic capability. In this study, optical microscopy is combined with computer vision image-based algorithms to characterize particles resulting from gear wear. Inspection at 50 × magnification captured approximately 70 % of the microplastic range, while complementary SEM analysis at 2500 × confirmed extended diagnostic capability at sub-micrometer scales. The circularity and aspect ratio of particles proved to be important indicators of wear stages, while the size of the particles remained consistent regardless of the torque applied. Approximately 80 % of the worn out volume was found consistently represented by 10 - 50 μ m particles. The developed methodology provides reproducible and quantitative insights into gear wear stages that are not accessible through conventional monitoring. The results demonstrate potential for improved monitoring and reliable failure prediction in polymer composite tribosystems.