search

Elsaam2y / DINet_optimized

An optimized pipeline for DINet reducing inference latency for up to 60% 🚀. Kudos for the authors of the original repo for this amazing work.
93 stars 15 forks source link

Next steps: Real time inference #15

Closed oijoijcoiejoijce closed 7 months ago

oijoijcoiejoijce commented 8 months ago

Any insight on how to speed up inference such that it runs in real time?

Elsaam2y commented 8 months ago

One possible way would be reducing the target resolution for the frames and mouth region, and retraining the model on this new resolution. This would speed up the inference significantly.