Hi BSTRobin, first of all, thank you for your feedback and suggestions. I took a closer look at both the code from the Github project and the AI Studio documentation. I see both as sensible and implemented as "prescribed". Unfortunately, I still didn't get a distinction between the two substances outside of AI Studio, so I recorded a data set directly above a container of compressor oil for a clearer distinction between compressed air and normal air (see picture). However, later I could not distinguish between normal air and the oil with the breakout board either (even though I could do this in AI Studio with an accuracy of 100 %, see screenshot). Therefore, I thought that oil might simply not be detectable and tried to distinguish between black tea and fruit tea in the next preliminary test. However, the result was the same. In the studio yes, but not possible with the breakout board. Even the simpler comparison of tea or normal air did not allow any differentiation outside the AI studio. To rule out a drift of the sensor on the breakout board compared to the Dev Kit, I also recorded the training data directly with the breakout board and transferred it to the AI Studio. However, the result was the same. Clearly distinguishable in the studio, but not in Live on the same sensor. The only thing I have been able to distinguish successfully and reliably so far was coffee from normal air. Therefore, my current status is that the sensor can only distinguish very strong fluctuations in the gas composition, odours (which tea actually also provides?). I therefore wonder how others can distinguish between different types of plastic or even fruit and assess the quality of a room's air with the sensor? I am currently at a loss and would appreciate any help. Thank you very much.
... View more
Hello everyone, I have the following experimental setup: I want to use the BME688 to distinguish between normal air and compressed air (containing oil). To do this, I have recorded 4 hours of training data in 2 sessions using the Development Kit. I then loaded the data into the BME AI Studio, processed it and trained an algorithm. The resulting algorithm had a Heater Profile with an accuracy of 99.75%. Extended testing of the algorithm in the AI studio on another set of training data (normal air) showed a clear assignment. In a further step, I now took the breakout board from pi3g.com and took the example project from https://github.com/pi3g/BME688CheeseMeatDetector and adapted it to my use case. If I now test my setup in normal air, the value runs unambiguously (1.0/0.0) to compressed air. Now I have the following questions: Has anyone had experience in this direction and can tell me whether the project is at all promising? What are the limits of the sensor in terms of detecting odours? What could be the reason that the test in the studio deviates so much from the application on the breakout board? Many thanks and best regards
... View more