Sightline Intelligence, a developer of video processing hardware and software solutions supporting target tracking and Intelligence, Surveillance, and Reconnaissance (ISR), has introduced its latest Software Release 3.10.2.
The update delivers advancements in real-time video processing, AI-powered detection, and edge-based ISR capabilities, alongside enhancements to tracking, geolocation, metadata, and camera integration.
Expanded AiTR Capabilities
The release brings full Aided Target Recognition (AiTR) capability to Sightline’s ultra-low SWaP 17xx platform, enabling advanced AI functionality on compact and embedded systems.
This expansion makes AiTR available across all Sightline and NVIDIA-based platforms, giving integrators greater deployment flexibility. The update supports real-time detection, classification, and tracking on low-SWaP systems, improves performance on NVIDIA platforms, and provides a consistent model and deployment workflow. As a result, the 1750-OEM is well suited for small Unmanned Aircraft Systems (sUAS), edge nodes, and compact payloads where size, weight, and power are critical.
Improved Detection Reliability with OOD Classification
Software Release 3.10.2 also introduces Out-of-Distribution (OOD) classification, a secondary validation layer that filters objects that do not match the trained AI model. This capability improves detection reliability, particularly in cluttered, unpredictable, and challenging environments.
OOD classification enhances accuracy, increases model trust, strengthens downstream tracking logic, and supports more confident mission-critical decision-making. As a result, operators can rely on AI systems that correctly identify valid targets while confidently rejecting detections that should not be classified.
AI Track Assist
The release introduces AI Track Assist, which combines Sightline’s tracking capability with AI detections to automatically initialize track boxes, adjust track size and position, and maintain tracking continuity when AI detections are unavailable.
This hybrid integration supports long-range surveillance with stable, low-drift tracking during motion, occlusion, and challenging visibility conditions, while reducing operator workload.
Advancements in AI, UI, Integration, & Geolocation
Software Release 3.10.2 introduces a new AI Model Loader user interface within Panel Plus, enabling direct loading of full-frame and classifier models. The updated interface simplifies model management and reduces setup time.
The release also expands geolocation and metadata functionality with support for terrain intersection calculations, altitude, frame-center, and slant-range metadata via KLV, along with enhanced VMTI classification for improved interoperability. In addition, the software enables the transmission of detection and tracking metadata only, reducing bandwidth requirements over constrained networks. These updates strengthen real-time intelligence delivery without requiring full video transmission.
A new ATAK plug-in enables direct operator connection and control of stabilization, image enhancement, and camera selection. This update improves the operator experience and supports broader open-source integration in future releases.
Software Release 3.10.2 supports the delivery of real-time intelligence at the edge, on the move, and in operational environments. Expanded AiTR capability, improved validation, and AI-assisted tracking contribute to more efficient, accurate, and mission-ready workflows.
