how to speed up the run time performance?

zhengqili / MegaDepth

Code of single-view depth prediction algorithm on Internet Photos described in "MegaDepth: Learning Single-View Depth Prediction from Internet Photos, Z. Li and N. Snavely, CVPR 2018".

MIT License

728 stars 169 forks source link

how to speed up the run time performance? #21

Open mmxuan18 opened 6 years ago

mmxuan18 commented 6 years ago

i try the model on my dataset, the result is magic. but in my nvidia 1070 gpu card, the python code time cost 170ms. did there anyway to speed up the performance?

kkk324 commented 6 years ago

You can reduce you input resolution to 320x240 or others. I have already tried on server, iPhone and Android and worked. But even the original size 384x512, single image will be 78 ms with 1080ti. 320x240 will be 35ms with same condition.

AlexandrGraschenkov commented 5 years ago

@CR-Ko how you convert it on mobile device? TF version + opencv::dnn?

l0stpenguin commented 5 years ago

@CR-Ko can you please share your converted model? Im trying to run this on ios but was not able to convert it.

l0stpenguin commented 5 years ago

@mlinxiang have you been able to improve the speed?