Gidel at Vision & AI 2025

Vision & AI 2025: Gidel Demonstrated FPGA Edge AI & Ultra-Fast High-Resolution Processing

At Vision & AI 2025, Gidel demonstrated two FPGA-based imaging workflows covering real-time Edge AI vision and ultra-fast high-resolution post-processing. The demonstrations featured FantoVision20-GigE with 10 GigE Vision acquisition, real-time HDR, and compression, together with SkyBoost for accelerating very large image-processing workloads.

FantoVision20-GigE: Real-Time FPGA Edge AI Vision

FantoVision20-GigE: NVIDIA Jetson + FPGA Edge AI Vision System

The live demonstration featured FantoVision20-GigE, combining NVIDIA Jetson Orin NX with FPGA-based 10 GigE Vision acquisition and real-time image processing.

The FPGA handled deterministic image acquisition, real-time HDR, image enhancement, and Quality+ Compression before transferring image data to the Jetson for AI workloads.

By moving these pixel-intensive operations to the FPGA, the system reduced acquisition and image-processing overhead on the Jetson. This left more CPU/GPU resources available for detection, tracking, classification, and application processing.

Engineers evaluating camera bandwidth, connectivity, and acquisition requirements can also review Gidel’s guide to GigE Vision cameras.

SkyBoost: Ultra-Fast High-Resolution FPGA Processing

SkyBoost: High-Resolution RAW-to-JPEG Imaging Acceleration

The second demonstration featured SkyBoost, Gidel’s FPGA accelerator for high-resolution image-processing workflows.

In the demonstrated workflow, SkyBoost reduced processing time from more than 20 seconds per image to more than two images per second. This dramatically accelerated post-processing workflows that could otherwise require many hours or even days.

Unlike the real-time FantoVision acquisition architecture, SkyBoost accelerated large image datasets inside a host workstation or server. This architecture supports inspection, aerial imaging, mapping, photogrammetry, analysis, and other high-resolution imaging workflows.

For a deeper look at this workflow, see Gidel’s article on aerial photogrammetry acceleration with SkyBoost.

FPGA Processing Across Real-Time and Post-Processing Workflows

The two Vision & AI 2025 demonstrations represented complementary stages of the imaging workflow.

FantoVision20-GigE demonstrated FPGA processing directly during image acquisition, including camera interfacing, HDR, compression, and data preparation before NVIDIA Jetson AI processing.

SkyBoost demonstrated FPGA acceleration after acquisition, processing very large image datasets inside a host workstation or server.

Together, these architectures show how FPGA processing can reduce CPU/GPU bottlenecks, improve deterministic performance, and accelerate imaging workloads across both real-time and offline applications.

Applications for FPGA Edge AI and High-Resolution Processing

The technologies demonstrated at Vision & AI 2025 support demanding imaging applications including:

• Industrial automation and inspection
• Embedded Edge AI
• High-speed machine vision
• Outdoor and rugged imaging
• Aerial imaging and mapping
• Multi-camera imaging
• High-resolution image processing
• Robotics and autonomous systems

Explore Gidel FPGA Imaging Solutions

• FantoVision Edge AI Systems
• SkyBoost High-Resolution Imaging Acceleration
• Real-Time HDR IP
• Quality+ Compression

Vision & AI 2025 Event Details

📆 Tuesday, December 2, 2025
📌 Pavilion 10, EXPO Tel Aviv

Official event page: Vision & AI 2025