Synthetic Drone Dataset (Experimental)

This repository contains an experimental synthetic dataset generated to demonstrate our ongoing work in synthetic thermal data for computer vision.
The current release includes synthetic thermal imagery of an FPV drone in simulated flight conditions. These assets are intended purely for research and demonstration purposes.

Important Notes

• This dataset is synthetic and does not represent real-world thermal conditions with guaranteed accuracy.
• No guarantees or warranties are provided regarding the quality, fidelity, or similarity of these images to real FPV drones or real-world environments.
• The dataset is provided as-is, without any commitment to performance or fitness for any particular application.
• Use of this dataset for safety-critical or operational purposes is strictly discouraged.

Purpose

Our goal is to showcase a developing technology for synthetic thermal data generation.
This dataset serves as an example of our ongoing research into physics-guided simulation of aerial thermal signatures and their application in computer vision tasks such as drone detection, tracking, and interception.

Dataset Overview

• Image count: 250
• Scene context: In-air only (no terrain or environmental background)
• Annotation format: YOLO (bounding boxes)
• Spectral domain: Thermal infrared (LWIR/MWIR simulated)
• Reference data: Derived from real-world thermal captures of an consumer-grade FPV drone, recorded using a FLIR 3.5 sensor as reference material

Status

This is an early, experimental release. Expect limitations, inconsistencies, and evolving formats as the project develops.

© 2025 SentIRX