Paddle-Anti-UAV

Anti-UAV base on PaddleDetection

Background

UAVs are very popular and we can see them in many public spaces, such as parks and playgrounds. Most people use UAVs for taking photos. However, many areas like airport forbiden UAVs since they are potentially dangerous. In this case, we need to detect the flying UAVs in these areas.

In this repository, we show how to train a detection model using PaddleDetection.

Data preparation

The dataset can be found here. We direcly download the test-dev split composed of 140 videos train the detection model.

• Download the test-dev dataset.
• Run unzip Anti_UAV_test_dev.zip -d Anti_UAV.
• Run python get_image_label.py. In this step, you may change the path to the videos and the value of interval.

After the above steps, you will get a MSCOCO-style datasst for object detection.

Install PaddleDetection

Please refer to this link.

We use python=3.7, Paddle=2.2.1, CUDA=10.2.

Train PP-YOLO

We use PP-YOLO as the detector.

• Run git clone https://github.com/PaddlePaddle/PaddleDetection.git. Note that you should finish this step when you install PaddleDetection.
• Move the anti-UAV dataset to dataset.
• Move anti_uav.yml to configs/datasets, move ppyolo_r50vd_dcn_1x_antiuav.yml to configs/ppyolo and move ppyolo_r50vd_dcn_antiuav.yml to configs/ppyolo/_base.
• Keep the value of anchors in configs/ppyolo/_base/ppyolo_reader.yml the same as ppyolo_r50vd_dcn_antiuav.yml.
• Run python -m paddle.distributed.launch --log_dir=./ppyolo_dygraph/ --gpus 0,1,2,3,4,5,6,7 tools/train.py -c configs/ppyolo/ppyolo_r50vd_dcn_1x_antiuav.yml &>ppyolo_dygraph.log 2>&1 &. Note that you may change the arguments, such as batch_size and gups.

Inference

Please refer to the infernce section on this webpage. You can just switch the configeration file and trained model to your own files.