Designing a reliable vision system for syringe testing — where every micron and millisecond matters
Precision Under Pressure: Building a Vision System for Auto-Injector Testing
Working with medical devices comes with a different kind of responsibility. You're not just solving a technical problem — you're working with tools that, one day, might save someone’s life. That reality sat with us from the moment a customer approached us with a challenge involving Auto-Injector.
They needed a system to test two critical parameters: needle penetration depth and drug flow time. Before anything else, they wanted a proof of concept (PoC) — something to show the approach was sound.
We started simple. A dummy syringe, a basic setup with a black matte background, a camera, and a light source. We manually triggered the syringe while capturing frames and used image processing techniques to estimate needle depth. Encouragingly, the results were consistent. The PoC did its job. The customer was satisfied with the progress, and we were confident in our foundation.
Then came the real test — the original Auto-Injector arrived.
Along with them came new requirements: live testing via a web application that could detect needle depth and flow time, display the results, and store them for later access.
We opted for a record-first, process-later approach. It allowed for a controlled environment where each test could be reviewed, frame by frame. But this is where the real engineering began.
The Auto-Injector needle, barely 0.2 mm in diameter, was almost invisible with our existing setup. The lighting wasn’t doing enough. So, we added a spotlight — only to realize that the background reflected just as much light as the needle. What we needed was separation, contrast, clarity.
Several experiments followed.
- We painted the background matte black
- Tried thin black paper
- Battled unwanted reflections
- Adjusted angles and intensity of lighting
Each attempt came close, but never quite nailed the result — until one combination worked: a ring light around the camera, paired with a matte black paper background. Suddenly, the needle was no longer lost in the noise. It was visible, trackable, and measurable.
We integrated this into the web application. The system could now:
- Record and analyze the needle’s motion
- Calculate penetration depth and flow duration
- Display real-time data and store test results
The customer was happy — but brought up a critical point.
User interaction. The person testing the Auto-Injector couldn’t have the camera in their way. So, we revisited the setup. After discussions, we agreed on a side-view camera angle, aligned at 90 degrees to the operator’s direction. It preserved usability without compromising accuracy.
In parallel, we also compared RGB vs monochrome cameras. Monochrome clearly offered better contrast and consistency for this application, so we recommended the switch — and got it approved.
Now, the system is live.
A custom web app captures every test. It shows penetration depth, flow time, and logs everything with full video evidence. The process is smooth, the data reliable, and most importantly — the customer is satisfied.
This project reminded us that even in the world of high precision and automation, success often lies in attention to small things — a reflection line, a camera angle, a lighting tweak. And when it comes to healthcare, those small things make all the difference.