reiinakano
commented
5 years ago Oh, cool, thanks for letting me know! Maybe I'll try porting the inception model too.
You should put up a tfjs demo so people can play around with it!
On Tue, Nov 20, 2018, 9:48 AM DavidGOrtega <notifications@github.com wrote:
Hi!
Great work!
I just only wanted to tell you that we just recently ported this model to tfjs from here https://github.com/elleryqueenhomels/arbitrary_style_transfer
We haven't modified the use of the InceptionV3. After the parameters optimisation the total model (we don't expose it as two parts) is just only about 25MB in total and the model is able to run (like yours) if the browser can handle enough RAM creating the WebGL context (usually limited to 512MB).
So your model is 17MB in total and the model with inceptionV3 is just only 25MB.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/reiinakano/arbitrary-image-stylization-tfjs/issues/3, or mute the thread https://github.com/notifications/unsubscribe-auth/ARg1Vog6e8ezIoie7Ywfhni9ZPaSXtWeks5uw1E_gaJpZM4YqKhh .
Hi!
Great work!
I just only wanted to tell you that we just recently ported this model to tfjs from here
We haven't modified the use of the InceptionV3. After the parameters optimisation the total model (we don't expose it as two parts) is just only about 25MB in total and the model is able to run (like yours) if the browser can handle enough RAM creating the WebGL context (usually limited to 512MB).
So your model is 17MB in total and the model with inceptionV3 is just only 25MB.
Of course that's not the REAL memory consumed by the model on inference.
Again, great work!