Hello! In this tutorial I will show you how to setup YOLO with Darknet.
YOLO (You Only Look Once) is a single stage detector that is popular due to it's speed and accuracy, it has been used in various applications to detect people, animals, traffic signals etc.
There are other object detectors such as R-CNN however, they are not as reliable as YOLO.
In this tutorial we will be building YOLOv3, v3 is significantly larger than previous models but it is the best one to use.
Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation.
First we need to install darknet
git clone https://github.com/pjreddie/darknet && cd darknet make
Next we need to download the pre-trained weights
wget https://pjreddie.com/media/files/yolov3.weights wget https://pjreddie.com/media/files/yolov3-tiny.weights
Or you can use the following bash script: https://gist.github.com/ethand91/62ec891bf30ba38ab47553e4952057e8
Now we can try out the detector on the sample images provided in the repository.
./darknet detector test cfg/coco.data cfg/yolov3.cfg yolov3.weights data/horses.jpg
If all goes well you should see the predictions in the terminal, there will also be a file called "predictions.jpg" which when opened should look something like the below image:
You can get the same results but with much faster time by enabling the GPU flag. To do this just modify the Makefile.
Then rerun make
In this tutorial I have shown how to setup Yolo with Darknet. I am currently interested in yolov4 👀
If you have any cool sample/projects please share. 😃
Like me work? I post about a variety of topics, if you would like to see more please like and follow me. Also I love coffee.