We tried to use all 4 classes in the BME AI-Studio:
Class A: Air
Class B: Coffee
Class C: Tea
Class D : Flower
Then in our application with BSEC Library we got for Air:
BSEC_OUTPUT_GAS_ESTIMATE_1=0.00
BSEC_OUTPUT_GAS_ESTIMATE_2=0.99
BSEC_OUTPUT_GAS_ESTIMATE_3=0.00
BSEC_OUTPUT_GAS_ESTIMATE_4=0.00
for Coffee the same:
BSEC_OUTPUT_GAS_ESTIMATE_1=0.00
BSEC_OUTPUT_GAS_ESTIMATE_2=0.99
BSEC_OUTPUT_GAS_ESTIMATE_3=0.00
BSEC_OUTPUT_GAS_ESTIMATE_4=0.00
It looks like BSEC_OUTPUT_GAS_ESTIMATE_1 doesn't change at all.
If we remove classes C and D from the algorithm, we get for Air:
BSEC_OUTPUT_GAS_ESTIMATE_1=0.99
BSEC_OUTPUT_GAS_ESTIMATE_2=0.00
and for Coffee:
BSEC_OUTPUT_GAS_ESTIMATE_1=0.00
BSEC_OUTPUT_GAS_ESTIMATE_2=0.99
this is correct.
BME AI Studio: 1.6.0
BME68x: 4.4.6
BSEC: 2.0.6.1
Solved! Go to Solution.
1. Sample rate must be BSEC_SAMPLE_RATE_HIGH_PERFORMANCE
2. Need to subscribe to all BSEC_OUTPUT_GAS_ESTIMATE_X (BSEC_OUTPUT_GAS_ESTIMATE_1,BSEC_OUTPUT_GAS_ESTIMATE_2,BSEC_OUTPUT_GAS_ESTIMATE_3,BSEC_OUTPUT_GAS_ESTIMATE_4). Even if you only use 2 classes.
You can see the logic of use in the example for BCEC 2.0.
Hello Danj,
Has you correctly labeled the gas? Maybe you can make the data collection time longer for each different kind of gas and have a try.
You shall make sure that the air is the fresh air without other smell.
Coffee for training and coffee for field test shall be the same concentration.
Hello Danj,
Could you give further feedback?