We make Smart Manufacturing a reality
Verifying consistently results of image-based read-outs
Artificial intelligence is frequently used to process, categorize and analyze images. Applications are broad and range from identifying cancers on CT scans to assigning a name to the face on a photograph or identifying handwriting – and many more.
While writing these algorithms is by now easy enough for even fairly unexperienced developers, simply being able to identify and/or categorizing pictures is not enough for industrial applications. The challenge lies in consistently verifying results of image-based read-outs in high throughput and even in situations where the images are of suboptimal quality, e.g. the image is not completely sharp, has optical distortions or the object that needs to be identified is shown at an angle.
Use Case: Diagnostic rapid test Results Audit
Verification of a visual read-out is important in medical diagnostics, e.g. when interpreting a diagnostic rapid test. These tests, which work like pregnancy tests were one line indicates a negative and two lines a positive result, are used to diagnose any number of infectious diseases such as HIV, Zika and – more recently – COVID-19.
While a healthcare professional or user administering the test can generally call the result themselves, AI-empowered results audit can verify the diagnosis within seconds by classifying the result based on a trained model. For this the users submits a picture of the device, e.g. using their smart phone.
This process can be done at scale, high speed and fully encrypted to comply with privacy laws. This step does not only confirm the result but also generates documentation for the physician to base their treatment decisions on.
Contact us to discuss how we can help you set up AI-empowered results audit.