(주)써모

 

Join us for an in-depth webinar on the technical challenges of Overhead Persistent Infrared (OPIR) surveillance and how synthetic imagery is enhancing deep learning performance for detecting ground-based targets from space-based platforms.

OPIR sensors aboard satellites offer powerful capabilities for Earth observation in the thermal infrared spectrum (MWIR and LWIR), but challenges like atmospheric interference and limited image resolution make it difficult to detect, differentiate, and identify ground targets with confidence. This session explores how synthetic EO/IR datasets—generated using MuSES and CoTherm—can supplement real-world data to improve training for machine learning algorithms in these demanding conditions.

We’ll present a case study using YOLO (“You Only Look Once”) deep learning models trained on synthetic datasets of adversarial ground vehicles across varying weather conditions, times of day, and operational states. The webinar will emphasize how image resolution impacts detection and recognition performance, offering insights into how future high-resolution space sensors might enhance OPIR effectiveness.

What You’ll Learn:

• Challenges in OPIR surveillance and data acquisition in the IR spectrum
• How synthetic data generation supports robust training for ML algorithms
• Use of MuSES and CoTherm to simulate realistic thermal IR overhead imagery
• Impact of image resolution on YOLO algorithm performance
• Implications for future space-based sensor system design

Who Should Attend:
 Professionals and researchers in remote sensing, aerospace defense, EO/IR imaging, and AI/ML applications in surveillance.

Explore how simulation, AI, and next-generation sensing platforms intersect to shape the future of global overhead surveillance.