Why a Plane Can Still Look Unknown

Introduction

In an automated instrumented UFO detector, aircraft are one of the largest sources of false alerts. The difficulty is that a distant airliner rarely resembles the neat side-on photographs used to train image-recognition systems. Instead, it may appear as a flashing point, an overexposed blob, a short streak, or a heat source only a few pixels across. For that reason, reliable aircraft filtering depends on combining visual evidence with independent flight information rather than asking a neural network to identify a tiny object from appearance alone. Multi-sensor confirmation—especially comparison against Automatic Dependent Surveillance–Broadcast (ADS-B) flight data—allows a system to distinguish ordinary air traffic from events that genuinely require further review.[MDPI]mdpi.comCommissioning an All-Sky Infrared Camera Array for…by L Domine · 2025 · Cited by 11 — Using a You Only Look Once (YOLO) machine le…

Why aircraft look strange in sky video

Machine vision has become remarkably capable at recognising objects in ordinary photographs, but sky-monitoring cameras operate under very different conditions.

An aircraft recorded by a fixed all-sky camera is often:

• only a few pixels wide;
• viewed from unusual angles;
• partially obscured by haze or cloud;
• saturated by sunlight or landing lights;
• interrupted by flashing anti-collision lights;
• distorted by atmospheric turbulence or camera motion.

These effects mean that two frames showing the same aircraft can appear dramatically different. A bright navigation light may dominate one frame, while the fuselage is barely visible in the next. Infrared cameras introduce another complication: engines and warm airframes produce thermal signatures that do not resemble conventional visible-light photographs.

Modern object detectors such as YOLO are designed to recognise patterns, but recognition accuracy inevitably declines as targets become smaller, noisier or more heavily distorted. Aircraft-detection research consistently shows that viewpoint, scale, background clutter and atmospheric effects remain important limitations even with advanced deep-learning models.[arXiv+2MDPI]arxiv.orgA Deep Comprehensive Review of Aircraft Detection…This paper critically evaluates and compares a suite of advanced object detecti…

For an automated UAP observatory, this means that the question is not simply, “Does this image look like a plane?” but, “Does every available source of evidence agree that this behaves like known aircraft traffic?”

How flight data helps confirm routine traffic

Visual appearance answers only part of that question. Aircraft filtering becomes much more reliable when the image is checked against independently measured flight information.

ADS-B is particularly valuable because many aircraft continuously broadcast their identity, GPS-derived position, altitude, heading and velocity. If a detected object occupies the same place in the sky at the same time as an ADS-B track, the software gains evidence from two independent sources rather than one. The FAA describes ADS-B as a surveillance system that provides precise aircraft position and identification information to both controllers and pilots.[Federal Aviation Administration]faa.govFederal Aviation AdministrationADS-B FAQ | Federal Aviation Administration3 Mar 2025 — With ADS-B, pilots can see what controllers see…

For automated UFO detection, the matching process typically works as follows:

This approach is considerably stronger than relying on appearance alone. Even when the aircraft itself occupies only a handful of pixels, its measured trajectory can still align closely with an independently reported flight path.

The Galileo Project has adopted this philosophy in its observatory design. Its published system combines YOLO object detection, SORT trajectory tracking and radio collection of ADS-B aircraft positions to establish a baseline census of ordinary aerial traffic before examining potentially anomalous events.[MDPI+2World Scientific]mdpi.comCommissioning an All-Sky Infrared Camera Array for…by L Domine · 2025 · Cited by 11 — Using a You Only Look Once (YOLO) machine le…

An important advantage is that trajectory consistency often proves more informative than image shape. A tiny flashing object moving at the speed and heading expected from a scheduled airliner is far less mysterious once its independently reported flight path matches the observed track.

Why image recognition alone produces false confidence

A neural network can assign a high probability to an incorrect label if it has limited visual information.

Several situations illustrate why appearance alone is insufficient:

None of these effects changes the underlying aircraft. They merely change the image reaching the camera.

Flight metadata provides an independent check that is unaffected by glare, blooming or camera exposure. When both visual tracking and flight telemetry agree, confidence increases substantially. When they disagree, the event is flagged for closer examination rather than immediately classified as anomalous.

This distinction is especially important in automated UAP systems because the objective is conservative filtering. Eliminating ordinary aircraft reliably is more valuable than forcing every detection into a visual category.

Where aircraft matching can still fail

Although ADS-B greatly improves aircraft filtering, it is not a perfect ground truth.

Not every aircraft broadcasts publicly available ADS-B information. Military operations, some government flights, certain historic aircraft and flights operating outside mandated airspace may not appear in public feeds. Coverage also varies with receiver networks, terrain and radio line of sight.[Federal Aviation Administration+2superiorskies.org]faa.govFederal Aviation AdministrationADS-B FAQ | Federal Aviation Administration3 Mar 2025 — With ADS-B, pilots can see what controllers see…

Timing and calibration also matter. Even a small clock error between the camera and flight-data receiver can produce an apparent positional mismatch. Likewise, inaccurate camera orientation or lens calibration can shift projected aircraft positions enough to prevent successful correlation.

Another limitation is that ADS-B itself should not be treated as infallible. Researchers have documented security and integrity concerns because the protocol was not originally designed with strong authentication. While this is primarily an aviation-security issue rather than a UAP issue, it reinforces the broader principle that robust systems compare multiple independent measurements instead of trusting any single source absolutely.[PMC]pmc.ncbi.nlm.nih.govPMCAutomatic dependent surveillance-broadcast (ADS-Bby W Ahmed · 2025 · Cited by 12 — Despite these advantages, ADS-B faces significant security vulnerabilities due to its open design an…

Consequently, well-designed observatories use flight databases as one layer within a broader evidence chain that also includes calibrated optics, trajectory reconstruction, timestamps and, where available, infrared, radio or other sensor modalities.

Why evidence fusion is the better filter

The practical lesson for automated instrumented UFO detectors is that aircraft filtering is fundamentally an evidence-fusion problem rather than an image-classification problem.

Image recognition answers, “What does this resemble?” Flight correlation answers, “Does this object behave like known air traffic?” Those questions are related but not identical.

By combining computer vision with calibrated tracking and independent flight data, automated systems can reject the vast majority of routine aircraft detections while leaving genuinely unmatched events for careful human analysis. That approach reduces false positives without claiming that appearance alone can determine whether an object is ordinary or genuinely unexplained.[MDPI+2World Scientific]mdpi.comCommissioning an All-Sky Infrared Camera Array for…by L Domine · 2025 · Cited by 11 — Using a You Only Look Once (YOLO) machine le…

Amazon book picks

Further Reading

Books and field guides related to Why a Plane Can Still Look Unknown. Use these as the next step if you want deeper reading beyond the article.

Book

Deep Learning

By Ian Goodfellow, Yoshua Bengio et al.

Rating: 3.5/5 from 6 Google Books ratings

Background for image classification models.

See on Amazon

Book

Artificial Intelligence

By Stuart Jonathan Russell, Peter Norvig et al.

Rating: 4.5/5 from 10 Google Books ratings

General AI concepts supporting automated classification.

See on Amazon

Book

Computer Vision

By Richard Szeliski

Explains recognition limits and object tracking.

See on Amazon

Book

Programming Computer Vision With Python

By Jan Erik, Jan Erik Solem

Demonstrates vision algorithms applicable to aircraft filtering.

See on Amazon

Browse more on Amazon: Deep Learning Artificial Intelligence Computer Vision

As an Amazon Associate I earn from qualifying purchases.

eBay marketplace picks

Marketplace Samples

Example marketplace items related to this page. Use the search link to explore similar finds on eBay.

Shop location

UsingUSA USA USAUKCanadaAustraliaIreland

Example eBay listing

Night Vision Monocular Digital 6 X Optical Zoom Records To TF Card To 300m Rang

SearcheBay.co.uk: night vision monocular

Browse similar oneBay.co.uk

Example eBay listing

Yukon Night Vision Monocular NVMT Spartan 4x50 Infrared with Case & Instructions

SearcheBay.co.uk: night vision monocular

Browse similar oneBay.co.uk

Example eBay listing

Infrared Night Vision Monocular 4X Digital Zoom Crosshair Scope HD Video Camera

SearcheBay.co.uk: night vision monocular

Browse similar oneBay.co.uk

Example eBay listing

Night Vision Binoculars Monocular Hunting Goggles Digital NV Camera Security/DVR

SearcheBay.co.uk: night vision monocular

Browse similar oneBay.co.uk

Browse more oneBay.co.uk

Example items shown for inspiration; availability and pricing can change. Branchoria may earn a commission if you purchase through outbound eBay links.

Endnotes

1. Source: mdpi.com
Link:https://www.mdpi.com/1424-8220/25/3/783

2. Source: faa.gov
Link:https://www.faa.gov/air_traffic/technology/adsb/faq

3. Source: arxiv.org
Link:https://arxiv.org/html/2404.02877v3

4. Source: mdpi.com
Link:https://www.mdpi.com/2072-4292/17/6/1001

5. Source: mdpi.com
Link:https://www.mdpi.com/1424-8220/25/10/3231

6. Source: faa.gov
Title: Federal Aviation Administration Automatic Dependent Surveillance
Link:https://www.faa.gov/about/office_org/headquarters_offices/avs/offices/afx/afs/afs400/afs410/ads-b

7. Source: superiorskies.org
Link:https://superiorskies.org/learn/adsb-technology

8. Source: pmc.ncbi.nlm.nih.gov
Title: PMCAutomatic dependent surveillance-broadcast (ADS-B
Link:https://pmc.ncbi.nlm.nih.gov/articles/PMC12192918/

9. Source: arxiv.org
Link:https://arxiv.org/html/2506.00125v1

Published: May 2025

10. Source: arxiv.org
Link:https://arxiv.org/html/2510.08333v1

11. Source: worldscientific.com
Link:https://www.worldscientific.com/doi/full/10.1142/S2251171723400081?srsltid=AfmBOooVr0ycY9BBUTECxXAVNtN4chaaAcaPDbXNdnQF-uwynmJhPyXN

Additional References

12. Source: avi-loeb.medium.com
Link:https://avi-loeb.medium.com/commissioning-data-on-half-a-million-objects-in-the-sky-from-the-galileo-project-observatory-are-a23bd084233a

13. Source: researchgate.net
Link:https://www.researchgate.net/publication/399827395_Web-Based_Aircraft_Detection_and_Tracking_System_Using_Computer_Vision_and_ADS-B_Data_Fusion_for_a_Resource-_Constrained_Airspace_Surveillance_System

14. Source: medium.com
Link:https://medium.com/%40angelinatsuboi/detecting-hacker-aircraft-using-artificial-intelligence-ec249baa866b

15. Source: researchgate.net
Link:https://www.researchgate.net/publication/355329242_Aircraft_recognition_from_remote_sensing_images_based_on_machine_vision

16. Source: youtube.com
Link:https://www.youtube.com/watch?v=uJtER5ahdPY

17. Source: leonarddavid.com
Title: Avi Loeb details the Galileo Project effort.Read more
Link:https://www.leonarddavid.com/unidentified-aerial-phenomena-research-paper-offers-insight-on-outing-human-bias-and-error/

18. Source: galileo.hsites.harvard.edu
Link:https://galileo.hsites.harvard.edu/publications

19. Source: aerialsoutheast.com
Title: ads b aerial photography
Link:https://aerialsoutheast.com/ads-b-aerial-photography/

20. Source: norma.ncirl.ie
Link:https://norma.ncirl.ie/7557/1/santoshkumarreddylekkala.pdf

21. Source: aerospace.honeywell.com
Title: Aerospace ADS-B Privacy
Link:https://aerospace.honeywell.com/us/en/about-us/news/2020/11/ads-b-privacy