November 10–13, 2026 | Messe München | Hall C5, Booth 331

At electronica 2026 in Munich, visitors to the Forlinx Embedded booth can try two camera-based workflows. In one demo, you type a description and find images captured earlier. In the other, you watch multiple video streams being analyzed side by side. Both run at the edge, using Forlinx modules based on Qualcomm Dragonwing IQ9 and IQ8 processors.

Alongside the live demonstrations, the Forlinx team met with Qualcomm representatives at the exhibition. The in-person exchange added an opportunity for direct dialogue as Forlinx presented its Qualcomm Dragonwing-based modules and edge AI demonstrations to international visitors.

Forlinx × Qualcomm at Electronica Munich Empowering Next-Gen Edge AI & Industrial Intelligence

Find Captured Images by Description on IQ9

Try typing “a person in a red shirt carrying a backpack.” The IQ-9075 demo searches a local image index and ranks the closest matches. You can see which images the system returns and change the wording to see how the results respond.

As the camera records, the system detects and tracks people, then uses SigLIP2 to encode images of those targets for a local index. A text query is encoded and compared with those features. The search covers targets already detected and indexed, and the results are ranked by similarity. Capture, indexing and search all run locally in the demo.

The corresponding Forlinx system-on-module (SoM) is the FET9075AA-C, based on Qualcomm Dragonwing IQ-9075. The configuration shown on the product card is rated at 100 dense TOPS and pairs eight Kryo Gen 6 performance cores with Adreno 663 graphics and up to 36 GB of LPDDR5 memory with inline ECC. The card lists support for up to 16 concurrent cameras and Linux options including Yocto and Ubuntu.

Figure 1. Forlinx FET9075AA-C product specifications.

Analyze Multiple Video Streams on IQ8

At the next demo, look for the tiled video display. The IQ-8275 system processes 36 video inputs, combining four live USB camera feeds and 32 local video streams. It decodes and prepares the streams for inference, then overlays detections on the corresponding tiles. Ask the booth team to enlarge a channel so you can see how an industrial safety model marks a result in the frame.

The corresponding Forlinx SoM is the FET8275AA-C, based on Qualcomm Dragonwing IQ-8275. The product card shows a 40 dense TOPS configuration with four Kryo Gen 6 performance cores, four efficiency cores, Adreno 623 graphics and up to 32 GB of LPDDR5x memory with inline ECC. It lists support for up to 12 concurrent cameras and Linux options including Yocto and Ubuntu. The 12-camera specification describes the platform’s camera capability. The 36 demo inputs describe the combined USB and local-video workload.

Figure 2. Forlinx FET8275AA-C product specifications.

Bring Your Vision Project to Booth 331

If your project calls for searching captured images, bring an example of what users would type and an estimate of how long you need to retain images. If your project calls for monitoring several video feeds, bring the input types, stream count, resolution, frame rate and models you plan to run. The Forlinx team can use those details to discuss what to test on your target configuration.

See the Forlinx event page to book a meeting or visit Hall C5, Booth 331 at Messe München.

---