Forum Replies Created
-
Posts
-
The reason for that is the size of the database.
I would recommend adding more samples and broadening your database, then evaluate again the model’s performance.
Ayelet
As I mentioned, since choosing the most successful pre-processing method is a trial and error process it is difficult to recommend using specific one. “Processed” is mainly useful for estimation models in which we assume the Beer-Lambert equation holds. “Normalized” is useful mainly for classification models. Sometimes use of both “processed” and “normalized” is needed when there is variance in the optical path of the samples.
We recommend trying each of the methods and evaluate the model’s performance.
Ayelet
Hi,
Below is a detailed explanation regarding spectroscopy and how SCiO works. I believe it will clarify most of the issues:
SCiO includes a light source that illuminates the sample and an optical sensor called a spectrometer that collects the light reflected from the sample.
The spectrometer breaks down the light to its spectrum (the spectra), which includes all the information required to detect the result of this interaction between the illuminated light and the molecules in the sample. This means that SCiO analyses the overall spectra that is received and, comparing it to different algorithms and information provided, identifies or evaluates it.
For example, if you know the basic spectra of a watermelon, and then see that as the watermelon gets sweeter, meaning it has more sugar content, the spectra gradually changes in a specific manner, you will be able to build an algorithm in accordance.
In recognizing the existence of a specific material, such as marijuana, in a sample, you will need to see if the reflectance of the material changes in a specific manner when the marijuana is present. Thus, you will need two samples of the material – with and without marijuana.
The basic idea is that we look at the overall reflectance and the why it changes – we do not separate and analyse it per specific molecules, rather we look at the whole composition.
This is why if your chemical composition varies between samples significantly and not only because of the factor you wish to analyse, you will need more samples, as with saliva.
We suggest you check out https://www.consumerphysics.com/myscio/technology and https://www.consumerphysics.com/myscio/images/dynamic/pdf/Spectroscopy.pdf for further information on spectroscopy.
Specifically, as for tomatoes, 40 is the amount of samples that we mention as a rule of thumb as a properly sized collection for a feasibility test. However, a comprehensive application should be based on hundreds of samples and thousands of scans.
Since SCiO applies machine learning algorithms, it should be noted that building the database for such an application requires not only access to samples but also their chemical characteristics.
This can be done either by us or by the SCiO developer community using our SCiO Development Toolkit (DevKit), which allows users to create models and apps. More info about the DevKit can be found here and It can be purchased here.
Let me know if you have further questions.
Ayelet,
The Consumer Physics Team
When using a prepared model, the result should be relatively consistent for different distances of the measure taken.
May I ask which pre-processing method did you use?
Generally speaking, using “processed and normalized” method is needed in order to eliminate differences in the optical paths.We have gone through significant engineering efforts to provide full backwards compatibility with the release of SCiO 1.1. This means that any model built with SCiO 1.0 is applicable to users of SCiO 1.1, with the same level of accuracy.
In other words: if you built a successful model with your SCiO 1.0 devkit, users of SCiO 1.1 will be able to use it. Your Devkit is not obsolete and can you continue to use it to build models which will serve the whole community of SCiO users: both 1.0 and 1.1.
However, Please note that only one version of SCiO can be used in a single collection when creating a model and collecting data – either 1.0 or 1.1.
Similarly, the liquid accessory should be used in a single collection when creating a model, however mixed collection with and without using the accessory is not recommended.
Ayelet,
The Consumer Physics Team
SCiO typically detects materials in concentrations of 0.1% or higher. Concentration levels of 0.1% or less may also be feasible for some materials, however this is very rare.
The exact specifications depend on the SCiO application being used and material being analyzed.
Ayelet
Hi,
The two versions differ by some of the hardware components. You can tell the difference by sight – 1.0 device has a temperature sensor in the optical head whereas 1.1 device has only one sensor, the illumination source.
The external temperature sensor does not perform well enough to use it for model creation – therefore it has been removed.
We have shipped SCiO v1.0 to developers. Currently, we only ship v1.1 devices.
Ayelet,
The Consumer Physics Team
Hi Guowei,
The Consumer app’s workshop applet allows you to work with the spectrometer by collecting material data and building analysis applets for your own materials. However, no customized pre-proccessing methods and algorithms are provided.
SCiO Lab offers various filtering and analysis tools which should be used in order to develop a successful and robust model.
According to our records, you have a developer license, hence using SCiO lab tools is highly recommended.
As for the questions:
1. Consistency is terms of measurement conditions is an important key factor for a successful data collection.
As you have suggested, using consistent amount of rice samples (or more precisely consistent layer depth) is recommended, so that the optical path of the reflected signal from each one of the rice samples will be identical.
2. Generally speaking, the differences between the spectrum curves should be visually inspected when reviewing the reflectance spectrum. Different pre-processing methods should be used: “Normalized” is useful mainly for classification models. Sometimes use of both “processed” and “normalized” is needed when there is a variance in the optical path of the samples.
Choosing the right pre-processing method is a trial and error process, it is recommended to try each one of the methods and evaluate the model’s performance accordingly.
I hope it was helpful.
Let me know if you have further questions.
Ayelet,
The Consumer Physics Team
Hi all,
We would like to clarify a few points that seem to have been misunderstood:
We have gone through significant engineering efforts to provide full backwards compatibility with the release of SCiO 1.1. This means that any model built with SCiO 1.0 is applicable to users of SCiO 1.1, with the same level of accuracy. In other words: if you built a successful model with your SCiO 1.0 devkit, users of SCiO 1.1 will be able to use it. Your Devkit is not obsolete and can you continue to use it to build models which will serve the whole community of SCiO users: both 1.0 and 1.1.
However, please note that only one version of SCiO can be used in a single collection when creating a model and collecting data – either 1.0 or 1.1.
We are committed to continually improving our products, and our customers expect us to continue innovating and delivering better products over time.
As we continue innovating, backward compatibility is very important to us. We will naturally strive to continue releasing improved versions of the SCiO, but as we do this we will continue to ensure that work already done by our developer community remains effective for newer models of the product.
Ayelet,
The Consumer Physics Team
Dear Alejandro,
If you or one of your colleagues would like to communicate in English, we will be more than happy to assist you.
Please contact support@consumerphysics.com with any question you may have.
Ayelet,
The Consumer Physics Team
The commentary refers to a liquid accessory prototype which was sent to a limited number of developers.
The production grade SCiO which will be shipped shortly, is an improved version of the prototype and will be provided along with an FDA approval for food safety.
Ayelet
Hi Ori,
Merging collections feature is in our roadmap, will be available in future versions.
Ayelet,
The Consumer Physics Team
Hi Ori,
This is great idea for SCiO’s next versions.
I’ll transfer your suggestion to our product manager for prioritization.
Ayelet,
The Consumer Physics Team
Hi Karly,
Seems like there is a hardware problem and the device should be replaced.
We will contact you privately via e-mail for further details.
Ayelet,
The Consumer Physics Team
