【Q&A】How do AI technology and industrial endoscopes work together?

Is AI defect recognition a defect recognition of pictures in the office?

Of course not, Evident's AI defect recognition is carried out in real time with the hole inspection at the inspection site, and it is a real-time video stream to look for defects, and it is of little significance if it is only looking for defects in the pictures that have been taken
.
If a customer wants to review in the office, they can simply keep the video and do it again
.

What types of AI defect recognition are now implemented?

AT PRESENT, OUR AI MODEL CONTAINS 11 TYPICAL DEFECTS: DENT/CRACK/NICK/TEAR/DEFORM/MISSINGMATERIAL/BURNS/BURNTHROUGH/TBCMISSING/MISSINGTIPCORNER/TIPCURL, AND ENGINEERS CAN CHOOSE WHICH DEFECTS TO SHOW ACCORDING TO THE ACTUAL SITUATION, SUCH AS WHEN INSPECTING THE COMBUSTION CHAMBER, YOU CAN CHOOSE TO ONLY SHOW CRACKS AND ABLATION, BURN-THROUGH, ETC.
And different color indications
will be displayed for defects of different severity.
Engineers can also adjust the confidence interval themselves to facilitate identification that is more user-friendly, in line with their own observation habits, and avoid losing small defects
.

Can blade counting identify forward and backward?

Of course, the blade counting function can actually automatically add and subtract according to the engineer's rotation direction, without manual intervention, and there are functions such as pause/start/end/hide in the software, which can be realized
with one click.

Is blade counting startup complicated to set up? Are there any requirements for viewing angles?

Evident's blade counting has two advantages: 1, the setup is fast and simple, and it can be started in seconds; 2.
Low requirements for the angle of the observer, even if the front and rear blades rotate at the same time in the field of view, there is no problem, and the effect is good
in LPC/HPC/HPT/LPT.
Expectably, we'll soon be adding a fully automatic feature in the next update, which doesn't require setup to start blade counting
.

(Operation Interface)

If there is shaking during observation, will it affect AI recognition and blade counting?

AI defect recognition and blade counting can be activated simultaneously at any time to record what defects
are on which blade.
Shabbing does not have any effect on AI recognition, because the advanced models we use can effectively avoid the negative effects
of shaking.
If the shaking is not strong, most of the inspected blades can still be seen in the whole picture, and it will not affect the blade count, and the accuracy is still guaranteed
.

Will there be a charge for software upgrades in the future? Can it be customized for customers?

The software was independently developed in China and is more responsive to the needs of Chinese customers
.
The AI model, functions, etc.
in the software will be continuously upgraded with customer needs to improve accuracy, and customers only need to purchase the software once, and they can enjoy free upgrades
in the future.

For more customers who want to train their own AI models according to their own requirements, combine endoscopes and engine management systems and quickly report on site, remote assistance, etc.
, we can provide customers with customized services, so that digital technology can help customers with hole inspection more efficient
.

What types of AI defect recognition are now implemented?

AT PRESENT, OUR AI MODEL CONTAINS 11 TYPICAL DEFECTS: DENT/CRACK/NICK/TEAR/DEFORM/MISSINGMATERIAL/BURNS/BURNTHROUGH/TBCMISSING/MISSINGTIPCORNER/TIPCURL, AND ENGINEERS CAN CHOOSE WHICH DEFECTS TO SHOW ACCORDING TO THE ACTUAL SITUATION, SUCH AS WHEN INSPECTING THE COMBUSTION CHAMBER, YOU CAN CHOOSE TO ONLY SHOW CRACKS AND ABLATION, BURN-THROUGH, ETC.
And different color indications
will be displayed for defects of different severity.
Engineers can also adjust the confidence interval themselves to facilitate identification that is more user-friendly, in line with their own observation habits, and avoid losing small defects
.