Article ID: 000091717 Content Type: Product Information & Documentation Last Reviewed: 09/01/2022

Inconsistent Results between Quantized Model in OpenVINO and Real-ESRGAN

BUILT IN - ARTICLE INTRO SECOND COMPONENT
Summary

Drawback of compressing model size

Description

Converted the REAL-ESRGAN model into OpenVINO IR (2021.4) and executed it.

Managed to run the IR but the result is different from REAL-ESRGAN.

Resolution

Compressing a full precision model into a smaller size format definitely has its own drawback.

The inferencing time would be faster but the trade-off is the accuracy.

If the use case requires to be accurate, such as clinical related results, it is not recommended to shrink the model as it will produce a less accurate prediction. Meanwhile, if the use case requires fast results without the need for prediction to be really accurate, then it is suitable to use a smaller size format.

Additional information

Choosing the right precision

Related Products

This article applies to 1 products