Code for the paper titled Lightweight Multi-Drone Detection and 3D-Localization using YOLO
-
Install the Ubuntu Linux distribution.
-
Open terminal and enter the following lines to build Darknet:
git clone https://github.com/pjreddie/darknet.git
cd darknet
make
Note: The folder data/dataset contains the image files and its corresponding annotation files in the same folder only. |
-
Move weights, data, scripts, cfg in the root directory of your cloned darknet.
-
Comment the corresponding cfg file you want to use, based on either test/train configuration as per mentioned in the comments of the cfg file.
-
Change lines 2 and 3 to your path in data/obj.data.
-
To do a train/test split, run the train_test_split.py in the scripts folder after changing the paths. The train.txt and test.txt files are generated inside the data folder.
To run prediction on an image named input.png (sourced inside the darknet root folder), open terminal in the root directory of the darknet executable and enter:
./darknet detector test data/obj.data cfg/tiny-yolov4_drone.cfg weights/backup_yolov4-tiny/yolov4-tiny_drone_best.weights input.png