Our Product InsightsPlatform Cameras

Unlock the power of Google Coral AI with e-con Systems™

Google Coral Dev Board is great for those looking to integrate AI and ML into the core of their vision applications. Find out why e-con Systems is your imaging partner of choice and how cameras like the e-CAM50_CUCRL help unlock the power of Google Coral AI.

power of Google Coral

As you may already know, the Google Coral Dev Board offers the complete toolkit to build AI product at the edge. You can take advantage of its on-device inferencing capabilities to create innovative vision models, along with a Tensor Processing Unit (TPU) that can perform 4 trillion operations per second (TOPS) while consuming low power. 

There’s no doubt that the Google Coral Dev Board is a valuable offering for those looking to integrate AI and ML into the core of their applications and products.

But today, for your vision-powered products to stand out in the market, you must cater to the niche expectations of your customers. While Google Coral Dev Board comes with exceptional inbuilt features, you need an experienced partner to maximize what it can do for your customers.

Why e-con Systems™ is your partner of choice

e-con Systems™, a leader in embedded camera solutions since 2003, offers cutting-edge camera support for the Google Coral Dev Board. Our powerful ISP makes it easy for you to develop and deploy edge AI products that are efficient, private, fast, and offline Our 5MP camera boards such as the e-CAM50_CUCRL come with the latest features that can handle large pixel sizes and capture high-quality images.

e-con Systems™’ long-standing commitment to covering all aspects of the embedded vision journey makes us a partner of choice. Our specialized expertise makes each step easier to take while helping you get the most out of your Google Coral Dev Board. 

From designing POCs (Proof of Concept), building unique carrier board for iMX8, and creating ML algorithms for specific data outputs to ensuring supply in large volumes and providing world-class support, we can boost your ability to leverage predictive maintenance, anomaly detection, machine vision, robotics, and a lot more.

Our cameras are also capable of streaming HD (1280 x 720) at 70 fps, FHD (1920 x 1080) at 60 fps, and 5 MP (2592 x 1944) at 25 fps in uncompressed (UYVY) format. The higher Signal to Noise Ratio (SNR) helps to produce clear images without noise and retain more details in shadows and highlights.

Key features

  • AR0521 CMOS Image sensor from ON Semiconductor
  • Low-noise for better inferencing 
  • High-resolution images for more details within AI ROI
  • 4- Lane MIPI for top-notch performance
  • V4L2 Framework for camera data access to applications through common frameworks as GStreamer

From manufacturing, healthcare, and retail to smart spaces and transportation, e-con Systems™ can unlock the true power of your Google Coral Dev Board.

Running inferencing samples on Google Coral with e-CAM50_CUCRL

e-CAM50_CUCRL is integrated to the Google Coral via the 4-Lane MIPI interface, which enables maximum throughput of the camera data. The V4L2 Framework native to Linux is used to capture camera frames from the CSI interface before sending it to the relevant application. Gstreamer framework ensures fast and easy access of image data at the application layers.

Google provides sample applications based on Python and GStreamer framework. The below link provides the application, as well as the steps to run real-time Inferencing – using a camera integrated to Google Coral Dev Board.

https://github.com/google-coral/examples-camera

The samples contain image classification and detection examples in real-time on the camera stream data.

image-classification-and-detection-using-e-CAM50_CUCRL

Google also offers samples based on different models trained on different datasets for easy integration and performance evaluation of the inferencing. Google Coral Toolkit gives support to tune the existing model with new samples. It can be done by retraining an existing model and compiling it to run on the EdgeTPU device.

Related posts

Leave a Comment