maghoumi / cudahog

An efficient implementation of HoG feature extractor in CUDA

Hi there 👋

My name is Mehran Maghoumi and I'm a senior deep learning engineer at NVIDIA. My primary area of work is parking space perception using surround camera setups for autonomous driving. I also hold a Ph.D. degree in computer sceince from the University of Central Florida. Feel free to checkout my full profile on my homepage.

What's all this? 🤔

Below is the list of my open source projects that ✨ I'm the most proud of ✨. I've worked on these either during my spare time or as a part of my Ph.D. dissertation. Countless hours of my time have gone into the development of each one, and nothing makes me happier than seeing people use them in their projects.

If you see something you like, please consider ⭐ starring ⭐ the repo. It gives me a better idea of where to focus my efforts!

Happy browsing! 💥

cudahog's Issues

Can not compile

/usr/local/cuda-7.0/bin/nvcc -I"/root/work/CudaHoG" -I/usr/local/include/ImageMagick-6 -I/usr/include/ImageMagick -I"/root/work/CudaHoG/include" -G -g -lineinfo -O0 --compile --relocatable-device-code=true -gencode arch=compute_20,code=compute_20 -gencode arch=compute_20,code=sm_21 -m64 -x cu -o "include/codefull/ImageUtils.o" "../include/codefull/ImageUtils.cu" -std=c++11
../include/codefull/ImageUtils.cu(60): error: no instance of overloaded function "codefull::FloatImage::convolve" matches the argument list
argument types are: (std::vector<Npp32f, std::allocator>, NppiSize)
object type is: codefull::FloatImage

../include/codefull/ImageUtils.cu(61): error: no instance of overloaded function "codefull::FloatImage::convolve" matches the argument list
argument types are: (std::vector<Npp32f, std::allocator>, NppiSize)
object type is: codefull::FloatImage