During that process, I read a bit about GPUs, CUDA and cuDNN.

The only problem I encounter was to update the NVIDIA driver, and it was done easy. It was not a painful experience(as I was expecting) to use this hardware because Udacity provided an AIM with the necessary software already installed, and I didn’t need to install anything else. darknet imtest data/eagle.As part of the Udacity’s Self-Driving Car Nanodegree, I had the opportunity to try a GPU-powered server for Traffic Sign Classifier and the Behavioral Cloning projects in Term 1. Then use the imtest routine to test image loading and displaying. You're done! To try it out, first re- make the project. Next, change the 2nd line of the Makefile to read: OPENCV=1 If you do this from source it will be long and complex so try to get a package manager to do it for you. If you want more support for weird formats (like CMYK jpegs, thanks Obama) you can use OpenCV instead! OpenCV also allows you to view images and detections without having to save them to disk.įirst install OpenCV. darknet -nogpu imagenet test cfg/alexnet.cfg alexnet.weightsĮnjoy your new, super fast neural networks! Compiling With OpenCVīy default, Darknet uses stb_image.h for image loading. If you compiled using CUDA but want to do CPU computation for whatever reason you can use -nogpu to use the CPU instead. darknet -i 1 imagenet test cfg/alexnet.cfg alexnet.weights If you want to change what card Darknet uses you can give it the optional command line flag -i, like. By default it will run the network on the 0th graphics card in your system (if you installed CUDA correctly you can list your graphics cards using nvidia-smi). Now you can make the project and CUDA will be enabled. Once you have CUDA installed, change the first line of the Makefile in the base directory to read: GPU=1 I won't go into CUDA installation in detail because it is terrifying. Compiling With CUDAĭarknet on the CPU is fast but it's like 500 times faster on GPU! You'll have to have an Nvidia GPU and you'll have to install CUDA. Great! Now check out the cool things you can do with darknet here. If you have any errors, try to fix them? If everything seems to have compiled correctly, try running it!. Gcc -I/usr/local/cuda/include/ -Wall -Wfatal-errors -Ofast -lm. Gcc -I/usr/local/cuda/include/ -Wall -Wfatal-errors -Ofast. If this works you should see a whole bunch of compiling information fly by: mkdir -p obj

If it doesn't work for you, email me or something? Installing The Base Systemįirst clone the Darknet git repository here. I've only tested this on Linux and Mac computers.

OpenCV if you want a wider variety of supported image types.īoth are optional so lets start by just installing the base system.

Darknet is easy to install with only two optional dependancies:

Should i install cuda on mac