winston1214 / Object_Detection_Drone

Person Detection in Drone Dataset using YOLOv5(contains UI Service)

Object Detection in Drone Dataset (Official Code in AI-Hub)

We released the dataset and utilization guide video on AI-Hub.

🎁 Full Version(Yolov5, mmdetection)

My Environment

• OS : Ubuntu 18.04.5
• GPU : NVIDIA RTX 2070 super
• CUDA : 10.2

Reference

• yolov5

• mmdetection

weight file

• yolov5 : weights/drone_survivor.pt 🥇
• cascade rcnn : https://drive.google.com/file/d/1mXANwNMbQU7tmmmhZ81dFaIRvXU6rDQL/view?usp=sharing
• faster rcnn : https://drive.google.com/file/d/1mXANwNMbQU7tmmmhZ81dFaIRvXU6rDQL/view?usp=sharing
• retina net : https://drive.google.com/file/d/1mXANwNMbQU7tmmmhZ81dFaIRvXU6rDQL/view?usp=sharing

How to do?

$ git clone https://github.com/winston1214/Object_Detection_Drone && cd Object_Detection_Drone

$ pip install -r requirements.txt

$ python3 ui2.py

More Details

Setting of input or output video location

Change this part

• Input Video location : Change 'source' option : ex. data/test.MP4 -> sample_dir/Sample.MP4
• Output Video location : Change 'project' option : ex. ui_test/ -> save_dir/

If you want to adjust the other options, you should refer here

Output

• Detection Results

• UI structure