1. Partium Help Center
  2. Partium Search Engine

What kind of part pictures can be used as reference images?

Partium works with a wide spectrum, from pictures taken with dedicated photo machines to smartphone-captured pictures and CAD renderings

Images created by a special photo machine

There are multiple types of purpose-built machines that can take pictures of an object from several angles automatically (“photo machines”). You need one user to place the objects accordingly and configure the machine to take the pictures. The resulting images can be used for machine learning with the Partium solution as well as for marketing purposes in case you want to sell those parts.

The images are often automatically segmented. This means, the background is replaced with a transparent alpha channel and only the object is visible in the image file. That has the advantage that different background conditions can be simulated during the machine learning process. 

The format of those images is usually .jpg if unsegmented or .png if segmented.

Images taken by an ordinary camera (smartphone, tablet or digital camera)

Those images are taken manually with an ordinary camera or even a smartphone. Those pictures require more effort to gather compared to a photo machine but are very helpful and even better suited for the machine learning algorithm. The reason is that users of the Partium App will find themselves taking pictures with similar quality making the training images closer to the real environment.

The format of such images is usually .jpg.

CAD renderings

CAD renderings are images of computer-generated visual representations of objects or parts created using Computer-Aided Design (CAD) software. Those images should aim for maximum detail and realism to enhance their utility in Partium's system, providing comprehensive visual data for accurate image recognition. Multiple angles of the part should be depicted to help in the training and recognition processes. 

The format of such images is usually .jpg.