Introduction

Anti-UAV tracking competition requires algorithms to track a given UAV target and simultaneously estimate the tracking states of the target. When the target disappears, an invisible mark of the target needs to be given.


News

🔥🔥🔥You can download test-challenge at Baidu Yun (y67q) / Google Drive🔥🔥🔥

🔥🔥🔥 Codalab server is online now!🔥🔥🔥


Guideline for Challenge

The dataset of test-challenge will be release on April 8th. The deadline for result submission is April 15th.

We build a baseline method for submit results on Github.

If you encounter any questions or misunderstandings, please feel free to contact us.

zhaojian90@u.nus.edu, qiang.wang@nlpr.ia.ac.cn

We also set up a QQ group (1013298408) for quick communication.


Participation requirements

The dataset of test-dev is NOT allowed to be used in learning.

Applying pretrained UAV detecter is NOT allowed.

Contestants are encouraged to submit previously published trackers or modified versions of third-party trackers.

The submission description should clearly state the algorithm framework.


Award

🎖️Best Paper Award: 500 USD + Certificate + XiaoMi Gift

🥇Challenge 1st-Place Award: 1200 USD + Certificate + XiaoMi Gift

🥈Challenge 2nd-Place Award: 800 USD + Certificate + XiaoMi Gift

🥉Challenge 3rd-Place Award: 500 USD + Certificate + XiaoMi Gift

Dataset

The dataset consists of 160 high quality, Full HD video sequences (both RGB and Thermal Infrared), spanning multiple occurrences of multi-scale UAVs.

Metrics

We define the tracking accuracy as:

acc = Σt(IoUt*δ(vt>0)+ pt(1-δ(vt>0)))/T

The IoU_i is Intersection over Union (IoU) between each corresponding ground truth and tracking boxes and the v are the visibility flags of the ground truth (the tracker’s predicted p are used to measure the state accuracy). The accuracy is averaged over all frames.

Results Format

For tracking with bounding boxes, please use the following format:

[

[x,y,width,height],

[x,y,width,height],

[],

[x,y,width,height]]

Note: box coordinates are floats measured from the top left image corner (and are 0-indexed) and empty list denote not exist flag.

Baseline and Evaluation Code

Baseline and Evaluation codes are available on the Anti-UAV github. Please refer to test.py.

We provide Thermal Infrared (IR), RGB videos and their ground-truth labels. Contestants can only use both IR and RGB videos and their ground-truth location in the first frame. Our evaluation ranks are calculated according to the results on the IR video.