[bug] gpu not at work while detecting

[simin75simin]

os: ubuntu 18.04
cmake instruction:

cmake .. -DCCTAG_WITH_CUDA:BOOL=ON

detection instruction:

./Linux-x86_64/detection -n 3 -i 0

the detection is rather slow.
and then i watched gpu usage during detection:

watch -n 1 nvidia-smi

but it's pretty close to 0% for the gpus.

help please?

p.s. multiprocessed detection in python
p.s. cmake output:

-- Build type not set, building in Release configuration
-- Try finding CUDA
-- BUILD_SHARED_LIBS ON
-- Found CUDA: /usr/local/cuda (found suitable version "10.1", minimum required is "9.0") 
-- Building in release mode
-- Found OpenCV: /usr/local (found version "3.4.1") found components: core videoio imgproc imgcodecs 
-- Found Eigen: version 3.3.9
-- CCTAG already there
-- BOOST already there
-- Found OpenCV: /usr/local (found version "3.4.1") found components: core videoio imgproc imgcodecs highgui 
-- TBB already there
-- DevIL found


******************************************
Building configuration:

-- CCTag version: 1.0.1
-- Build type: Release
-- Build Shared libs: ON
-- Build applications: ON
-- Build tests: ON
-- Build documentation: OFF
-- Cuda support: ON
-- Compiling for CUDA CCs: -gencode;arch=compute_35,code=sm_35;-gencode;arch=compute_37,code=sm_37;-gencode;arch=compute_50,code=sm_50;-gencode;arch=compute_52,code=sm_52;-gencode;arch=compute_60,code=sm_60;-gencode;arch=compute_61,code=sm_61;-gencode;arch=compute_70,code=sm_70;-gencode;arch=compute_75,code=sm_75
-- Disable Eigen alignment: ON
-- Enable AVX2 optimizations: OFF
-- [debug] Serialize all the output: OFF
-- [debug] Enable visual debug: OFF
-- [debug] Disable output stream: ON
-- [debug] nvcc additional warnings: OFF
-- Install path: /usr/local

******************************************


-- Configuring done
-- Generating done
-- Build files have been written to: /home/lty/fusion/cctag_gpu_build/CCTag/build

[anand97]

Try running it with the --use-cuda option
i.e. ./Linux-x86_64/detection -n 3 -i 0 --use-cuda

[simin75simin]

Try running it with the --use-cuda option i.e. ./Linux-x86_64/detection -n 3 -i 0 --use-cuda

I am no longer working on this project but thanks for the response.