Back to All
Partner Blog

Real-Time AI Inspection in a Compact System-on-Module for Automated, AI-powered Quality Inspections using Visual Analytics

Qualcomm Dragonwing™ QCS6490 Helps Enable On-Device Inference and High-Speed Camera Integration for Compact Industrial Vision Systems

Customer
European Industrial OEM

Solution Partner
SECO

Industry
Industrial Manufacturing

Application
2D Industrial Vision System / Quality Inspection

Qualcomm Technology
Dragonwing QCS6490

Featured Offering
SECO SOM-SMARC-QCS6490

Overview

A European industrial OEM was developing a compact 2D industrial vision system for automated quality inspection. The system needed to capture high-resolution images, identify defects immediately, and operate continuously inside a sealed enclosure.

Processing inspection data within the system would remove the need for an external industrial PC or cloud connection. The challenge was delivering sufficient AI and vision performance within strict space, power, and thermal limits.

The OEM engaged SECO, a leading provider of edge computing and AI solutions for industrial applications, to design an embedded computing platform and provide integration support. SECO selected the Dragonwing QCS6490 processor and integrated it into its SOM-SMARC-QCS6490 module.

The resulting platform helped enable local inspection, direct camera connectivity, fanless operation, and Linux-based development while reducing the custom hardware and software work required to reach production.

AI at the edge

The Challenge: Balance AI throughput, camera processing, and thermal limits in a sealed industrial design

The customer needed to combine high-resolution image acquisition and continuous AI inference in a design built for factory-floor operation.

Key requirements included:

  • On-device defect detection: AI inference needed to run within the sensor so inspection decisions could be made without depending on external processing or network connectivity
  • Fanless operation: The system had to operate inside a fully enclosed, compact housing with minimal thermal dissipation, making fanless operation a hard design constraint
  • High-speed vision integration: Smooth support for high-resolution camera inputs was required for the vision sensor’s image acquisition and AI-based inspection pipeline
  • Industrial-grade reliability: Continuous 24/7 operation in harsh manufacturing environments demanded long-term hardware stability and wide operating temperature support
  • Long-term platform scalability: Linux ecosystem support and sufficient compute headroom were required to evolve AI algorithms over the product lifecycle without a full hardware redesign

The Solution: A SMARC module for AI and vision processing based on Dragonwing QCS6490

SECO supported the project with its SECO SOM-SMARC-QCS6490 based on the Dragonwing QCS6490 processor. The SoC is designed to deliver a heterogeneous architecture that distributes AI, vision, and compute workloads across dedicated engines simultaneously. SECO packages the Dragonwing QCS6490 into a production-ready 82 × 50 mm SMARC® 2.1.1 module with 8 GB LPDDR5, optional UFS 3.1 drive up to 128 GB, high-speed interfaces, and full OS support, giving the customer a production-ready compute platform that helps reduce hardware integration effort significantly, and accelerates time-to-market.

The solution includes:

  • 12 dense TOPS on-device AI: The integrated Qualcomm AI Engine helps enable real-time computer vision inference to run locally on the sensor, eliminating cloud dependency and processing latency
  • Efficient heterogeneous processing: The Qualcomm® Kryo™ 670 CPU and Qualcomm® Adreno™ 643L GPU are built to support concurrent application, image analysis, and vision workloads within the system power envelope
  • GPU-accelerated image analysis: The Adreno 643L GPU is engineered to accelerate computer vision and image analysis workloads, helping enable advanced inspection algorithms without external graphics hardware
  • Integrated vision subsystem: 2x MIPI-CSI 4-lane camera interfaces with a dedicated ISP and is designed to deliver smooth, high-quality camera connectivity directly from the SMARC® 2.1.1 module
  • Production-ready embedded platform: SECO’s 82 × 50 mm module with Yocto Linux support, combined with SECO’s embedded software ecosystem is structured to accelerate integration and reduce time-to-deployment
  • Scalable SMARC® architecture: The standardized SMARC® 2.1.1 form factor and SECO's long-term product availability aim to provide a clear migration path to future compute generations, allowing customers to preserve carrier board investments, extend product lifecycles, and reduce redesign effort
Qualcomm-image
Figure 1: SECO SMARC® 2.1.1 module powered by Dragonwing QCS6490 processor

Results

  • Real-time on-device inspection: AI-based defect detection is structured to run within the sensor, eliminating external PC and cloud processing.
  • Fanless compact design: Achieved the required AI and vision performance within the enclosure’s power and thermal limits.
  • Simplified system architecture: Integrated processing, acceleration, and camera support helped reduce component count and design complexity.
  • Faster development: SECO’s production-ready SMARC module helped reduce custom hardware, integration, and software bring-up.
  • Reliable continuous operation: The platform is designed to support sustained use in industrial manufacturing environments.
  • Scalable AI capabilities: Available compute performance and Linux support allow inspection models to evolve without immediate hardware redesign.

Opinions expressed in the content posted here are the personal opinions of the original authors, and do not necessarily reflect those of Qualcomm Incorporated or its subsidiaries ("Qualcomm"). The content is provided for informational purposes only and is not meant to be an endorsement or representation by Qualcomm or any other party. This site may also provide links or references to non-Qualcomm sites and resources. Qualcomm makes no representations, warranties, or other commitments whatsoever about any non-Qualcomm sites or third-party resources that may be referenced, accessible from, or linked to this site.

© Qualcomm Technologies, Inc. and/or its affiliated companies.

Snapdragon and Qualcomm branded products are products of Qualcomm Technologies, Inc. and/or its subsidiaries. Qualcomm patents are licensed by Qualcomm Incorporated.

Note: Certain services and materials may require you to accept additional terms and conditions before accessing or using those items.

References to "Qualcomm" may mean Qualcomm Incorporated, or subsidiaries or business units within the Qualcomm corporate structure, as applicable.

Qualcomm Incorporated includes our licensing business, QTL, and the vast majority of our patent portfolio. Qualcomm Technologies, Inc., a subsidiary of Qualcomm Incorporated, operates, along with its subsidiaries, substantially all of our engineering, research and development functions, and substantially all of our products and services businesses, including our QCT semiconductor business.

Materials that are as of a specific date, including but not limited to press releases, presentations, blog posts and webcasts, may have been superseded by subsequent events or disclosures.

Nothing in these materials is an offer to sell or license any of the services or materials referenced herein.