Open BigPeng opened 4 months ago
Very good work. Could you extract the loading of the three models of facedetection, face_restore_model, and swap_model and load them in advance, batch processing of images will be much faster.
ReActor loads them during the first queue's run and after that all the processing goes very fast
Feature description
Very good work. Could you extract the loading of the three models of facedetection, face_restore_model, and swap_model and load them in advance, batch processing of images will be much faster.