Live data from Hacker News

Launch HN: Cerrion (YC S22) – Computer vision to reduce production line problems

cerrion.com

11–20 of 26 posts

Re: Launch HN: Cerrion (YC S22) – Computer vision to reduce production line problems

#11
post #10

Nice product. May I ask the differentiator with commercial packages for industrial vision like Halcon or Cognex?

Thank you very much. Current industrial vision solutions like Halcon or Cognex focus on quality control i.e they inspect products for defects while we focus on process control where we look for anomalies in the process itself i.e in the material flow, material handling etc.

Re: Launch HN: Cerrion (YC S22) – Computer vision to reduce production line problems

#12
Would be curious if you are able to integrate metrology products that generate point clouds vs video feeds. I have a buddy that is in that space for automotive manufacturing, they use a variety of laser scanning products for production line setup/calibration/etc but a runtime monitoring system that can detect small deviations might be an entirely new revenue stream.

Re: Launch HN: Cerrion (YC S22) – Computer vision to reduce production line problems

#14
post #13

For anomaly detection, does your team have to worry about concept drift and data distribution shifts? If so, how do you combat that?

Great question. Yes, that's a big topic for us. We closely monitor production metrics to detect things like input changes and output drift. This allows us to automatically detect if something changes, so that we can trigger retraining and redeployment.

Re: Launch HN: Cerrion (YC S22) – Computer vision to reduce production line problems

#15
post #13

For anomaly detection, does your team have to worry about concept drift and data distribution shifts? If so, how do you combat that?

Great question. Yes, that's a big topic for us. We closely monitor production metrics to detect things like input changes and output drift. This allows us to automatically detect if something changes, so that we can trigger retraining and redeployment.

We built a lot of this in-house though, as we didn't find suitable existing tooling for that

Re: Launch HN: Cerrion (YC S22) – Computer vision to reduce production line problems

#16
post #12

Would be curious if you are able to integrate metrology products that generate point clouds vs video feeds. I have a buddy that is in that space for automotive manufacturing, they use a variety of laser scanning products for production line setup/calibration/etc but a runtime monitoring system that can detect small deviations might be an entirely new revenue stream.

We're not doing anything which is not video-based at the moment as we rely on versatility of optical sensors. However you are totally right regarding the value of precise point cloud generation — definitely something to look at in the future.

Re: Launch HN: Cerrion (YC S22) – Computer vision to reduce production line problems

#18
post #9

Earlier quoted context omitted.

I'd suggest not doing this. What you're doing is not novel. You have a much better chance if you focus e.g. food and beverage. There's many reasons but your setup will go smoother, GTM will be easier and you will get higher NPS. You will also have the opportunity to develop novel aspects.

We are indeed focusing - my comments regarding other areas where my personal thoughts on the above question and not Cerrion's strategy.

I was going to say this seems best as a packaged solution + consulting/setup fee. I could see this working in fast-food assembly stations as well, but may have more variance with human element.

Also could add on an analysis component of "study my flow and quantify potential jam points".

Re: Launch HN: Cerrion (YC S22) – Computer vision to reduce production line problems

#19

What camera hardware are you using? https://www.luxonis.com/ might be an interesting option.

We use off-the-shelf cameras (we just need an RTSP stream), but thanks for the input, we'll definitely check them out!
Post reply on HN