szagoruyko / imagine-nn

IMAGINE torch neural network routines
Other
109 stars 35 forks source link

no CUDA_KERNEL_LOOP in ROIPooling #24

Open szagoruyko opened 8 years ago

szagoruyko commented 8 years ago

As spotted by @gidariss there is no need in CUDA_KERNEL_LOOP in the kernels as it's executed for every point of output. small change, but I want to keep ROIPooling clean and easy to understand as it's used and referenced by many people.