February 15, 2016 at 5:04 pm #2648sakrelaastaParticipant
I have been working (playing) with my SCIO for some time now, and I have gathered some feedback. I thought to send it with a email, but then I decided that is better to have it on the forum, so more people can read, and add.
Generally the 3 first are the most important (from my perspective), and then there is no specific order, just some notes/ideas. Also: I use firefox in windows and the scio lab app is on android (if this is needed). I attached some simple pictures, not something special, but I believe that they help to present the matter.Where I mention “user”, I mean the non-developer user. Where I mention “app”, I mean the app that a user will use, and not the scio lab mobile.
1. Vessel for scanning liquids.
Taken care! Good work!
2. Statistical information
Have been mentioned before. After a scan, it would be nice if the app, if with the result of the model, it would also give a percentage of certainty for the result. Especially for the qualitative analysis, it is very important to know if the result was only 4% mach. Another idea in this area is to give more than one result. For example if I scan a cheese and want to see what kind of milk it has, it will be very nice if the result was 38% cow, 12% sheep (the percentages are not concentrations, but certainties), instead of just “cow milk”. That way I can understand that, first of all, it isn’t cow milk, and that maybe it is a combination of these two milks.
3. Offline use
– For the developers. Probably the easiest one. Just store the data on the device, and when I am online again, I press a button to upload them. For my case, that I want to go to quarries and usually there is no cell connection, means that I cannot use it.
– For the users. This is a little more difficult, because all the data/models etc must be downloaded to the device. But again, what to do with an app that can recognize mushrooms or herbs, if I cannot use it on the forest or mountain?
4. Maybe the models can be programmed for special circumstances/results. For example, if a result for a value of concentration is negative, just give “below detection limit” or “error”. It would be nice also to provide with the results, the limits of the model. For example “fat 23% (1%-65%)”. Where the numbers in the parenthesis are the two limits.
5. The indicator for the scio battery on the scio lab mobile doesn’t work correctly. Before the update was always at 100%. Now it is reducing, but again it is at 78%, then I turn off the device, and when I turn it on it is on 95%! And 2-3 times it was turned off because of the battery, while the scio app mobile was showing that it was above 50%.
6. During the scanning, the “Quick scan mode” is very nice addition. But maybe it could appear also in the page with the spectrum (after a scan). In most cases with a new item, I want to see the results for the first scan, but then I would like to be able to “fast scan” the repetitions. Maybe also add a “make x automatic continuous scans” after the first scan. For me that I use it in a stand and for every sample I make 6 scans, it would be very nice to be able to make, after the first scan, the other 5 automatically.
7. Being able to choose some scans or samples from a collection and copy or move to another collection.
8. At the spectrum tab, to have on top of the table some numbers of how many samples and scans are chosen in every case (see picture 1 – a)
9. At the model tab, the table for qualitative analysis is not very… nice (picture 2). The words in the lines are not correctly aligned. And the words in the columns are cut. Maybe should have the whole name, or when we pass with the mouse over a name, the full name could appear. Moreover with more samples, the area of the table doesn’t increase, resulting to cramped tables (picture 3).
10. At the model tab, when I choose a point, it would be nice if it would give me some information. Maybe the number of the scan, so to be easier to find a point that is very wrong for the model. Or even better, all the information, like when we choose a scan on the spectrum.
11. The only way I have found to delete a scan is to choose it on the spectrum tab, then click on one scan on the diagram, then “details” and then “delete”. Is there an easier way? Also it would be very nice if we could delete many scans faster. Maybe checking boxes before each scan (or group of scans) that we can choose and then delete all them together? For example if I wanted to permanently delete all the sheep milk (picture 1 –b).
12. The “boxes” of the 11 comment could be used also for faster choose of scans. If we want “some of that” and “some of these”, instead of making many filters.
13. When I change a wavelength on a spectrum, the part that is outside the new wavelength is now grey. If I change the wavelength to the previous value, the grey part doesn’t change.
14. The spectrum tab, changes on the default all the time. For example in the “hard cheese” collection, if I choose “analyze by fat” and then I decide to add a filter “milk type = cow”, then it returns to the default “analyze by brand”.
15. The same as above with the scan tab. If I choose to present the results by dissenting concentration of fat, and then add/remove a filter, it returns to the default arrangement.
16. At the scan tab, the bar to move down at the table is cut/not appearing (see picture 4). Also if I have many attributes in my collection, and therefore many columns on the table, there is no horizontal bar, to move further right and I can move only by choosing a cell and the pressing “enter”.
17. I know that automatically, in every scan the temperature, the time and the location are stored. I can see the temperature and the time, in the scio lab, but I cannot find the location data.
18. It would be nice if when we zoom on a spectrum, could be possible to “move” on the spectrum and to “follow” the lines. Maybe by holding down the mouse button and moving it.
19. It would be nice in a collection, if we could make some “if the first model gives xxx then run also the 3rd model”. Example: Let’s say I have made an app about milk. When a user runs the app, first there is a model deciding if the milk is from animal or plant. It decides that it is “animal”. Then a new model specific for the animal milk runs, and decides that it is “cow milk”. Now two more models run and the one calculates fat with the model created only for “cow milk”, and another model decides if it is fresh, condensed, pasteurized or high-temperature pasteurized milk. If I had different models for each case (cow milk, goat milk, coconut milk, almond milk etc.), there would be many models running after each scan. If it was possible to add an algorithm “if the first result is that, then use that model”, the used models for each scan would be much fewer.
These few notes for now! I know they appear as a lot, but some are simple things, some are just ideas. The great job that has been done with the scio and the lab excites me, so I am trying to help with feedback as much as I can.
- This topic was modified 7 years, 7 months ago by sakrelaasta.
Attachments:You must be logged in to view attached files.February 16, 2016 at 6:06 am #email@example.comParticipant
Great ideas. I have seem questions\feedback.
2. You are talking about user applications, ones that are written by us, the developers. I did not try to write an app yet, but is it not entirely up to us to add it into our application? Or is it impossible and you need the SCiO developers to pitch in?
3. Yeah, offline raw-storing of data would be great. On the user’s side of the application, I think it’s still up to us to add it to our application. But for raw data scanned by us the developers, having the ability to scan and save the raw data for later analysis would be great!
4. Again, isn’t it only up to us?
5. Did not run into that yet (well, I did not scan a lot yet…). Is it common for others?February 16, 2016 at 8:43 am #2666sakrelaastaParticipant
2.3.4. I also haven’t created an app yet, but with the use of the “test model” and as I understand it, no, it not up to us. The models and equations are property of the company, and we don’t have access to them. We give the data, and the “cloud” creates the equations, stores them on the cloud and give back only statistical information about the model itself (like R^2). When someone uses the model, it just gives the result. The result is just “what comes closer from all the possible answers”, so even if you scan gasoline and run it by a model for milk, it will give you the closer answer between “cow”, “goat” or “sheep” milk even if the best answer is a 2% mach. That is also why we also see negative concentrations.
And because the models are property of the company, it isn’t easy (or maybe it is impossible) to be downloaded for offline use.
On the other hand, yes, the limits of a method are known to us, because we give the data. So yes, that is up to us.
If I have understand it wrong, please correct me!
* All the discussion concerns simple users and developers. I have the impression that researchers have access to more (or all the) information.
note: Don’t make me wrong. I am exited with scio, but I look how it would be better, and what makes me reluctant to create an app for users.
February 22, 2016 at 3:15 pm #2685cro55anParticipant
- This reply was modified 7 years, 7 months ago by sakrelaasta.
Im really interested in the overall hardware and would love to get feedback from the developers seeing as I cant get my hands on a scanner 🙁 .
From your experience, how is its design, capabilities and overall ease of use, where you impressed by its functionality. Are the results you receive effective & efficient. Did you find that the ambient light plays a major role in the effect of your results. I find its hard to specify a main application for the device, do you find yourself looking at it(applications) from the perspective of developing specifically for home users/consumer or targeting industrial applications? Can you see environmental uses, and suppose I would really like to know if it exceeded expectations?
Any feedback would be greatly appreciated
Cro55anMarch 15, 2016 at 11:23 am #2965Igor NOVICKSpectator
Very useful read, indeed.
- You must be logged in to reply to this topic.