An overview of Picsellia, an MLOps platform for computer vision.

I was able to attend a demo of this product recently, and it warrants your interest. I am not affiliated with Picsellia, the goal is to allow people to discover this product and the MLOps practices it allows.

Picsellia offers a one-of-a-kind product. After HuggingFace, we could be in front of the second Unicorn in the field of MLOps platforms.

As HuggingFace was able to succeed in asserting itself in the field of Natural Language Processing, Picsellia puts forward a rather impressive set of tools to manage MLOps on Computer Vision projects, from A to Z.

For the moment Picsellia is only accessible in its paid version. We would have liked to see a model hub or a community data hub, as these functions are available with the tool.

The company was founded a couple of years ago and the product is already mature. The database hosts 10 million images.

Let me show you around!

Data management

All images are duplicated and organized by metadata. We can access them via the graphical interface or through queries in…