top of page

Core AI Technology

Large Language Model as Copilot

Our production copilot leverages the latest large language model technology to meet your specific production needs. Simply integrate your data, and our AI is ready to assist you in your day-to-day activity.

See our use cases on how Copilot can assist your production today.

Vision-based object detection, segmentation, and counting

We developed an AI model that allows you to detect your unique products. Once detected, we keep track of your production by counting the number of products being produced throughout the day.  Here is an example from one of our clients site where they use our AI model to detect and count the number of products being produced every day.

odSeg.png
Screenshot 2022-07-21 162341.png
vlcsnap-2024-01-25-16h05m54s548.png

Vision-based defect/anomaly detection

We also developed AI models to detect defects in your production line. In addition, these models can detect these defects even when the products are running on a fast moving conveyor line. 

We present a case study where a client uses our Laplace One vision and robotic platform to perform defect detection and rejection on a fast moving conveyor. The study demonstrates how our model can detect these defects in real-time, which then allows our robot to separate these defects from the line.

Vision-based keypoint detection and pose estimation

In some cases, we might want to detect specific parts of an object, and maybe its orientation. For example, this can then be used in conjunction with robotic solution that allows the robot to pickup an object at a specific location and orientation with precision.

kpd_viz_2 (1).jpg
image.png

AI Demo Hands-On

Text-guided vision

In this demo, we showcase a text-guided object detection model. Given a user query (e.g., "candy cane") and image, the model will detect the queried object within the image. The model can detect vast variation of different objects, and we provided some example images for you to get started with, or you can also opt to test the model with your image. While this particular model may not be able to immediately detect some very specific objects that only your facility produces, our AI model can easily be finetuned to detect these. Learn how you can use our AI model in your facility today.

Text-guided vision + anomaly detection

In this demo, we show how we are able to leverage the latest technology in text guided segmentation model to help us solve real-world defect/anomaly detection application. Similar to our text-guided vision demo, given a user query and an image, the model can locate the object of interest within the image. However, in this particular demo, we have finetuned this model to perform complex defect/anomaly detection task for a specific candy production line.

bottom of page