Seven Segment AI Camera Athena - InCloud株式会社

Seven Segment AI Camera: Athena

Our Athena system is an AI-powered seven segment digital display detection system. The AI model is running on Tensorflow Lite and can be run entirely offline on a Raspberry Pi or any other Single Board Computer. The system can be connected via HTTP to our inbuilt GUI or via UDP/HTTP API to your own application.



Speed and accuracy

7-segment display detection is done within 5ms for up to 4 digits with 99% accuracy under ideal conditions. With additional training for your particular placement conditions, we can train the model for just as fast and accurate results.



Offline mode

Our Tensorflow Lite model can be run completely offline meaning devices that run our Athena software can be placed in areas without internet connectivity. This provides a lot of flexibility in places that the devices can be placed. For example, factories with lots of wireless interference can be a strong use case.



double shutter logic

Our software does not rely entirely on the AI model for accuracy. We have also included many other logic functions for increased accuracy and result confirmation. Our double shutter logic compares two photos taken within milliseconds apart to confirm the result accuracy.

Client

Internal Product

Location

Tokyo, Japan

Technology

Tensorflow Lite
Raspberry Pi 4
USB Camera

Challenge & Solution

7-segment digital displays are used in various types of electronics, especially, in scenarios where IoT connectivity can be very difficult to implement due to lack of internet connection, wireless interference, or hardware limitations. However, gathering accurate data from these displays can be easily done using our Athena software.

Our Athena system offers extremely accurate, lightning fast detection, and completely offline solution for detecting 7-segment digital displays.

99.8% accurate and 5ms detection rate under ideal conditions. Can be trained for various angles and lighting conditions.

Online and offline modes

Completely customizable solution based on your use case. We can customize and train the model for your hardware or any other requirements that you may have.