Yolo it's cool, but installing it's a pain. Which OpenCV, which CUDA? This script will make your life a bit easier.
Docker is the most beautiful thing in the world: you can install anything you want and it will work everywhere (yep, even Darknet).
- You have both CPU and GPU verisons.
- Python example for yolov3-tiny (weights included)
- All config and data for darknet are included (if something is missing go here)
https://docs.docker.com/install/
Then
sudo bash build_cpu.sh
https://github.com/NVIDIA/nvidia-docker
Then
sudo bash build.sh
On my computer (powered by i7 5820k) it took almost an hour.
Because you have to enable desktop forward, so run xhost +
on your host (not the container).
Edit build_darknet/Makefile at row 50-51 and specify your CUDA path.
Open the Dockerfile and search for
# Install python dependencies
#RUN apt-get install python3-tk -y
#RUN pip3 install matplotlib
Then jst edit anything as you would do in a native machine (not as on Anaconda).