Growing Image and Audio based AI applications with neural network (NN) and machine learning algorithm processing at the edge devices has increased need for AI Computing in low power embedded devices in Smart Home, Consumer, Industrial, Automotive and Healthcare industry. The shift of AI to the edge from the cloud helps faster real-time decisions at edge devices. Our edge AI based AICD (Audio and Image Classification Device) reference design helps our customers to build smart edge AI IOT devices.

Client challenges

In centralised IoT architecture, all the sensor data is processed at the cloud and requires a huge amount of data transfer. The image and audio based AI processing based smart devices requires send large amount of data to cloud and thus increases data transfer bandwidth, cost and huge cloud computing resources and energy consumption. Bringing AI functions to edge devices for real-time decision avoids sending all these data to the cloud. Over the past few year “deep learning” gained momentum and thus increased client’s challenges on how to optimally use deep learning on big data to extract meaningful business values. Deep learning is an approach to AI (Artificial Intelligence) and ML (Machine Learning). ML is more often described as sub-discipline of AI. AI and ML are showing great promise when it comes to developing the autonomous, self-teaching systems which are revolutionizing many industries.

With edge AI processing, the data is processed faster, and a real-time decision is taken in the user environment. Consumer privacy and security of data is also taken care of edge AI. As data is processed at the edge, the privacy information of consumers is not being transmitted to the centralized infrastructure, the risk of misuse can be reduced and thus increases security of data.

Eoxys AI Based AICD Device

AI based AICD (Audio and Image Classification Device) combines both video and voice data by running both AI image and audio processing algorithms. This approach is called multimodal AI and with the help of the device becomes more contextually aware and aware of individual behaviour.
  • ARM Cortex M4 MCU

  • Audio AIML processor with 256 layers CNN

  • Image AIML processor with 64 layers CNN

  • 4 Mics for best Audio classification applications.

  • VGA (640x480) 1/6” image sensor with options for RGB or IR 840nm image capture

  • WiFi 2.4GHz a/b/g

  • Rechargeable battery (2400mAH)

Solutions we can build

With our edge AI based AICD (Audio and Image Classification Device) reference design, we can build following helps our customers to build smart edge AI IOT devices with reduced power consumption, reduced cost and faster release of products, and also helps in directly processing at the edge device without cloud computing.

Vision based AI applications.
  • With edge AI Camera, Industrial manufacturing factories can effectively use AI functions for visual inspection for defects, robotic control for assembly and many other vision use cases.

  • With AI Camera trained on CNN, the smart home appliances like microwave oven recognizes and identifies the food and its cooking conditions and thus users can get various cooking recipe suggestions and overcook warning based on food placed in it.

  • With AI Camera in smart refrigerator, the refrigerator can use Food image classification AI processing to identify the quality of food available in the refrigerator.

  • AI Video analytics-enabled security cameras can help to detect important activities such as detect a person than an object and real-time face recognition from moving images.

  • The AI enabled smart camera offers promising applications for elderly and baby monitoring for their caretakers. This camera can monitor sleeping condition of a baby or an elderly persons.

  • The AI enabled camera helps to analyse fall detection, loitering detection, occupancy detection, people counting and restricted areas access detection.

Voice based AI applications.
  • Natural Language Processing (NLP) with maximum number (30 to 50 keywords) of speech keywords detection for multi-lingual languages at the edge devices without cloud server.  

  • AI voice commands classification integration for Smart TV applications.

  • The voice assistants with local AI Audio classifications, can provide better human interface and thus personalize the device to match with the their preferences.

  • AI audio analytics based device can detect and alert the sound of a baby crying, glass breakage, gunshots, and other malicious sounds.