Added a 5 point face landmarking model that is over 10x smaller than the 68 point model, runs faster, and works with both HOG and CNN generated ...
確定! 回上一頁