search

MaybeShewill-CV / bisenetv2-tensorflow

Unofficial tensorflow implementation of real-time scene image segmentation model "BiSeNet V2: Bilateral Network with Guided Aggregation for Real-time Semantic Segmentation"
https://maybeshewill-cv.github.io/bisenetv2-tensorflow/
MIT License
224 stars 59 forks source link

A question about GFLops #24

Closed Flyzanetothesky closed 4 years ago

Flyzanetothesky commented 4 years ago

the GFLops of Paper is 21.15 with the input of spatial size is 2048 x 1024, but why your GLops is 34.53 ??? @MaybeShewill-CV

Screenshot from 2020-09-15 14-56-17

MaybeShewill-CV commented 4 years ago

@Flyzanetothesky Maybe some difference between my implementation and official implementation:)

Flyzanetothesky commented 4 years ago

@Flyzanetothesky Maybe some difference between my implementation and official implementation:)

thanks for your quickly response, I wonder your input size of model when you calculate GFLops , 2048x1024 or others ? @MaybeShewill-CV

MaybeShewill-CV commented 4 years ago

@Flyzanetothesky https://github.com/MaybeShewill-CV/bisenetv2-tensorflow/blob/34ba8e7e939efa387188510f9eb74cd27f260a8a/tools/cityscapes/timeprofile_cityscapes_bisenetv2.py#L291-L302 Graphy's input tensor size was [512, 1024]