This workshop aims to promote research and development of the state-of-the-art Computer Vision-based models in the domain of autonomous flight safety. Despite significant advances in Computer Vision, tracking small objects, such as those found in Airborne Objects Tracking (AOT) dataset, is still challenging. As opposed to the traditional Multi-Object Tracking (MOT) datasets, AOT dataset poses a unique combination of challenges, due to the tiny size of the airborne objects and real-time performance requirements. Autonomous drones engage Sense and Avoid (SAA) technology for collision avoidance from unforeseen airborne obstacles on their planned route. Computer Vision algorithms for detection and tracking of airborne objects with low-cost, lightweight cameras onboard these drones will allow to scale safe autonomous flight.

Please visit this website for more details about Airborne Object Tracking challenge.

The challenge is over - we THANK all the participants for the great results and CONGRADULATE Dmytro Poplavskiy for winning the first and creativity prizes!!!

Start time (AM - EST)

07:30 : Workshop schedule by Yuri Federigi

07:40 : Hosting An Object Tracking Challenge - How hard can it be ? by AIcrowd

07:55 : Buffer / Q&A

08:00 : Video-Based Aircraft Detection for Collision Avoidance Purposes by Pascal Fua

08:35 : Buffer / Q&A

08:40 : A New Tracking-by-Attention Paradigm by Laura Leal-Taixé

09:15 : Buffer / Q&A

09:20 : Segmenting and Tracking Every Pixel by Andreas Geiger

09:55 : Buffer / Q&A

10:00 : Introduction to Sense and Avoid Systems in Autonomous Drones by Amir Navot

10:35 : Buffer / Q&A

10:40 : SIAM-MOT Baseline of the AOT Challenge by Bing Shuai

10:55 : Buffer / Q&A

11:00 : Winning Solution of the AOT Challenge by Dmytro Poplavskiy

11:20 : Runner-up Solution of the AOT Challenge by Yulu Gao

11:40 : Generating Synthetic Images and Labels for Airborne Objects by Franco Callari

11:50 : AI Assisted Annotation for Airborne Objects by Chengxi Ye

12:00 : Q&A