Gidel at IMVC 2026

IMVC 2026: Gidel Demonstrated Real-Time FPGA Imaging and Edge AI

At IMVC 2026, Gidel demonstrated two live FPGA-based imaging workflows covering real-time acquisition, embedded AI, and high-resolution image processing. The demonstrations showed how FPGA acceleration can reduce CPU/GPU workload, improve deterministic image processing, and support demanding machine vision applications.

Real-Time Embedded Imaging & AI Solution

FantoVision40-CXP12: Mini Edge AI System with FPGA

The FantoVision40-CXP12 demo presented a complete embedded AI vision pipeline combining FPGA-based image acquisition and processing with NVIDIA Jetson computing. The FPGA handled deterministic acquisition from multiple sensors, performed real-time preprocessing, and applied Gidel’s proprietary Quality+ Compression to reduce data volume before the Jetson handled AI workloads such as tracking, detection, and classification.

The system acquired data in real time from two CoaXPress-12 cameras, one IR and one RGB Bayer, processing 13MP images at 30 FPS while maintaining high image quality.

Engineers evaluating CoaXPress camera selection can also review Gidel’s guide to CoaXPress cameras to compare bandwidth, link count, cabling, synchronization, and acquisition requirements.

High-Resolution Post-Processing Acceleration

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

The second demonstration featured SkyBoost, Gidel’s FPGA-based imaging accelerator for very large images, including image sizes above 100MP.

The demo used a PCIe FPGA accelerator installed in the host PC, with image processing running directly on the FPGA. This offloaded the main imaging workload from the host CPU or GPU and accelerated high-resolution post-processing. The same FPGA image processing pipeline can also be deployed on Gidel’s FantoVision system when a compact, stand-alone embedded solution is required.

In the demonstrated workflow, SkyBoost reduced the processing time from 30 seconds or more per image to more than two images per second. This can significantly shorten high-resolution processing workflows that otherwise require hours or, in some cases, days.

The technology accelerates analysis, reporting, and decision-making in demanding applications. These include defense, intelligence, aerial imaging, and mission-critical vision systems.

SkyBoost supports host-based post-processing workflows, while SkyBoost-RT extends the FPGA processing approach to real-time image processing during acquisition.

FPGA Frame Grabbers for CoaXPress, GigE Vision, and Camera Link

Gidel also presented FPGA-based acquisition platforms for demanding machine vision and imaging systems. Gidel PCIe frame grabbers combine high-bandwidth camera acquisition with programmable FPGA resources for optional real-time image processing, image enhancement, compression, data reduction, and application-specific processing.

Solutions included CoaXPress frame grabbers, GigE Vision frame grabbers, and the Camera Link frame grabber.

FPGA Acceleration Across the Imaging Workflow

The two IMVC 2026 demonstrations represented different stages of the imaging pipeline. FantoVision40-CXP12 showed how FPGA processing can operate directly at acquisition, handling camera input, preprocessing, compression, and data preparation before NVIDIA Jetson AI processing.

SkyBoost demonstrated the complementary post-processing architecture, where FPGA hardware accelerates very large image datasets inside a host workstation or server.

Together, these architectures show how Gidel FPGA platforms can offload demanding imaging tasks from the CPU and GPU, improve throughput, reduce system bottlenecks, and maintain predictable processing performance across both real-time and offline workflows.

IMVC 2026 Event Details

📆 Monday, June 29, 2026
📌 Pavilion 10, Booth 9, EXPO Tel Aviv

Event link: IMVC 2026