- Create the train-test split
python3 utils/splitTrainAndTest.py ../../database/train
- Install Darknet and compile it.
cd ~
git clone https://github.com/AlexeyAB/darknet.git
cd darknet
make
- Get the pretrained model
wget https://pjreddie.com/media/files/darknet53.conv.74 -O path/to/darknet/darknet53.conv.74
-
Fill in correct paths in the darknet.data file, Fill the full path!.
-
Start the training as below, by giving the correct paths to all the files being used as arguments
cd ~/darknet\
./darknet detector train path/to/CC_Detection/train/darknet.data path/to/CC_Detection/train/darknet-yolov3.cfg ./darknet53.conv.74 -map
The train will take a few days, you can see at the buttom of the graph how many hours remain.
- file yolo-obj_xxxx.weights will be saved to the /path/to/CC_Detection/train/weights for each 1000 iterations
- file yolo-obj_last.weights will be saved to the /path/to/CC_Detection/train/weights for each 100 iterations
If you need any help please contact me - shira.levi11@gmail.com