Introduce your business and what you do there.
I’m the VP of technology at a company that manufactures hardware and software related to medical cannabis use. We have software that’s designed to help people manage chronic pain.
OPPORTUNITY / CHALLENGE
What challenges were you trying to address with ThirdEye Data?
We had a machine learning project that we wanted to get going to see if we could use it as a basis for some of the later machine learning projects that we’d like to do as part of our software platform offering.
What was the scope of their involvement?
Part of what our software does is helping chronic pain sufferers to input their current medications. Since medication names are often difficult, we wanted to skip that process of typing out the name. We wanted to help users automatically determine what kind of medication and dosage they’re using based on a photograph of their medication, which is uploaded to our servers. Then, we’d come back with an informed guess for what that medication is.
We knew the project would involve machine learning. ThirdEye Data put together a proposal that suggested a couple of different approaches, and we narrowed it down to one of them. We decided to go with TensorFlow, along with some optical character recognition (OCR) libraries and some Fuzzy Matching strips behind the scenes that could take an approximation of what someone had typed or from a photograph of an actual pharmaceutical label.
Sometimes the OCR isn’t 100% perfect, and so the Fuzzy Matching lets us take something that had either been typed or had been taken from an image and cross checks it against our list of 5,000 known medications for pain.
We’re currently working with them on a different project.
What is the team composition?
There were six or seven team members involved, including a project manager.
How did you come to work with ThirdEye Data?
I wasn’t involved in selecting ThirdEye Data. We had a couple of conversations, and they walked us through some of the projects that they had done in the past that were similar to what we were looking to do. They put together a proposal for us, and we decided to go ahead with that pilot project.
How much have you invested with them?
We spent $25,000 for this project.
What is the status of this engagement?
We started working together in February 2018, and the work lasted a month.
RESULTS & FEEDBACK
What evidence can you share that demonstrates the impact of the engagement?
One of the main outcomes of this project is that we realized the large amounts of data that we would need to successfully run future machine learning projects. There’s a public pill dataset composed of images, and ThirdEye Data did a great job of multiplying those images to do the machine learning training.
To productize something like that, we would need many more images. In terms of our direction, the project let us know what we would need to do to make something like that successful in a consumer-facing product.
However, we’ve taken pieces of the technology they developed and deployed it into our product, especially the Fuzzy Matching stuff that they did.
How did ThirdEye Data perform from a project management standpoint?
We had weekly or biweekly meetings to run through the project’s status. Their communication was fantastic, and documentation and the code delivery were top notch. We were able to selectively take parts of what they had done and deploy it with other teams without any issues.
Furthermore, the quality of their work was terrific. Any time there were any issues or new lessons that had come out of what they’d done, we could easily recalibrate what we were doing.
What did you find most impressive about them?
The speed at which they completed the work was impressive, and their cost was good in terms of value.
Are there any areas they could improve?
No, I can’t name anything.