False post-process value witch "extended Resolution" false

[haverland]

The Problem

The first digit will on value >9.5 to 0, if extended resolution is set false.

In second part of the condition should be (result_after_decimal_point >= (1-Digital_Uncertainty) * 10)

AI-on-the-edge-device/code/components/jomjol_flowcontroll/ClassFlowCNNGeneral.cpp

Line 146 in 09ecd72

if ((result_after_decimal_point <= Digital_Uncertainty * 10) || (result_after_decimal_point >= Digital_Uncertainty * 10)) // Band around the digit --> Rounding, as digit reaches inaccuracy in the frame

But I am wondering why the first digit must be handled in transition. Where's no need for it.

Version

15.2.4

Logfile

[0d08h38m12s] 2023-06-07T06:45:01     <DBG>   [CNN] PointerEvalHybridNew - No predecessor - Result = 0 number: 9.500000 number_of_predecessors = -1.000000 eval_predecessors = -1 Digital_Uncertainty = 0.200000
[0d08h38m12s] 2023-06-07T06:45:01     <DBG>   [CNN] getReadout(dig100)  prev=0
[0d08h38m12s] 2023-06-07T06:45:01     <DBG>   [CNN] PointerEvalHybridNew - NO analogue predecessor, zero crossing has taken placen = 2 number: 2.500000 number_of_predecessors = 9.500000 eval_predecessors = 0 Digital_Uncertainty = 0.200000
[0d08h38m12s] 2023-06-07T06:45:01     <DBG>   [CNN] getReadout#PointerEvalHybridNew()= 2
[0d08h38m12s] 2023-06-07T06:45:01     <DBG>   [CNN] getReadout#result= 20
[0d08h38m12s] 2023-06-07T06:45:01     <DBG>   [CNN] PointerEvalHybridNew - NO analogue predecessor, no change of digits, as pre-decimal point far enough away = 0 number: 0.000000 number_of_predecessors = 2.500000 eval_predecessors = 2 Digital_Uncertainty = 0.200000
[0d08h38m12s] 2023-06-07T06:45:01     <DBG>   [CNN] getReadout#PointerEvalHybridNew()= 0
[0d08h38m12s] 2023-06-07T06:45:01     <DBG>   [CNN] getReadout#result= 020

Expected Behavior

Result of the first digit should be the int-value of the recognition.

Screenshots

No response

Additional Context

No response