The paper that presents this algorithm has the best known accuracy on the widerface dataset, which is why it is called state of the art. The authors of the pape

The paper that presents this algorithm has the best known accuracy on the widerface dataset, which is why it is called state of the art. The authors of the paper published an implementation of this algorithm based on mxnet, but a lot people and companies use tensorflow instead of mxnet in their work, so just using the mxnet implemenation is not an option. Thats why I converted it to TF, with a slight decrease in accuracy on widerface. Then what is "acceptable" depends on your judgement, but the widerface dataset is extremely challenging, with many pictures having hundreds of small faces. In some of these pictures my implementation will miss a few faces, or find them but with lower probability. Overall for the vast majority of face detection applications, the two implementations will yield extremely similar results. I havent yet found a picture with a few dozen faces where the TF implementation performed not as well as the original one !