We present AssistTaxi, a robust dataset specifically curated to address the critical need for high-quality data in the field. AssistTaxi is designed to capture real-world scenarios and encompasses data collected from Melbourne Orlando International Airport (KMLB) and Grant-Valkaria (X59) general aviation airport.
The AssistTaxi dataset has been recorded from a Piper Cherokee Warrior, single-engine, fixed gear aircraft. Three GoPro Hero 8 cameras were fitted in the cockpit having camera angles of the center, right and left. This was done after discussions with aircraft experts to adjust camera angles and positions. The camera has a 12MP sensor with capabilities of recording videos in 4K/60p, 2.7K/120p and 1080/240p modes (up to 100 Mbps bit rate) utilizing its HyperSmooth 2.0 video stabilization feature. The video format can be in both H.264 and H.265 codecs. Another set of data was collected with the GoPro Hero 10 camera, with a 23MP 1/2.3-inch sensor,GP2 Processor, capable of recording 5k/60 fps, 4k/120 fps, 2.7k/240 fps videos. The data was collected from taxiing and multiple takeoff landing operations at the Melbourne Orlando International airport (KMLB) and Valkaria airports (X59). KMLB, is a Class D general aviation airport, 1.5 miles northwest of downtown Melbourne in Brevard County, Florida, and 50 miles southeast of Orlando, on Florida's Space Coast. X59, is a Class G general aviation airport, located 1 mile west of the central business district of the city of Grant-Valkaria in Brevard County, Florida. A total of 15 videos were recorded and then frames were extracted resulting in 350,000 images. The frame extraction was carried out using a Linux system, which had the x86_64 architecture and an Intel Core i7-9700K CPU running at 3.60GHz. The system had a total of 15GB of RAM and was running Ubuntu 22.04.1 LTS.