UAV detection under yolov8
The dataset is avaliable in https://github.com/Jake-WU/Det-Fly.
Fyi, the dataset need to be splited manually
This is the code for UAV detection, and implementing Loss Landscape in YOLOv8.
This is an unclean code for generating a loss landscape. You need to modify some code in your scenario.
If the code does not work due to version update, here are some following steps you could reproduce the code:
<1> modified the validators -> get the loss value
<2> run model.val -> get the direction dictionary
<3> run the main.py in the file PlotLoss.
https://github.com/ultralytics/ultralytics
If you use our implementation of Loss Landscape in YOLOv8
plz cite our work, thanks!
@INPROCEEDINGS{10507293,
author={Lai, Hawking and Liu, Bowie and Kan, Ho Yin and Lam, Chan-Tong and Im, Sio Kei},
booktitle={2023 9th International Conference on Computer and Communications (ICCC)},
title={YOLOv8-lite: An Interpretable Lightweight Object Detector for Real-Time UAV Detection},
year={2023},
volume={},
number={},
pages={1707-1713},
keywords={Convolutional codes;Privacy;Convolution;Atmospheric modeling;Decision making;Machine learning;Detectors;Object detection;UAV detection;YOLOv8;Depthwise convolution;Interpretable machine learning},
doi={10.1109/ICCC59590.2023.10507293}}