OTDAU now includes tools for online and offline analysis of collected data. Typically, a long data collection run will record many event sequences triggered by mundane targets, such as aircraft and birds. Manual analysis of those events requires opening each Group folder and playing at least one recorded file for the event, looking for true unknowns. That can take a long time if many events are recorded.
OTDAU Data Analytics can greatly reduce the time to analyze collected data by automating recognition of known target objects in collected data. It utilizes advanced methods of machine vision and deep learning technologies to scan all of the files under a user-selected folder, determining the most likely initial target object in each and then modifying the associated folder name to include its identification and confidence measure. The user can then focus further analysis on files labeled UNKNOWN or those with low confidence. The identification process may be run online, following each target event or offline, for a batch of events under a single folder.
Currently, objects that the software attempts to recognize include:
In daytime lighting conditions —
• Aircraft of all types as well as helicopters
• Birds in many modes of flight
• Foliage such as trees
In nighttime conditions –
• Aircraft with standard blinking navigation lighting
Please see the latest revisions to the UFODAS User Guide, V1.07, downloadable from ufodap.com.