Machine learning (ML) and its benefits to a product's software suite can be significant, but most engineers today aren't as in command of the techniques and approaches needed to create something that can provide a value-add. Developing a ML solution can often be challenging, especially if the tools used aren't intuitive or easy to implement. Addressing this need, Imagimob, an Infineon Technologies company, is updating its Imagimob Studio.
Now, users can visualize their ML modeling workflows and develop better edge device models faster. This latest release includes a major user experience upgrade, the Graph UX interface. Designed to bring greater ease and clarity to the ML modeling process, it also offers advanced capabilities such as built-in data collection and real-time model evaluation.
“What we’ve done here is represent ML projects in a way that makes more sense to users and gives them more freedom to operate,” said Alexander Samuelsson, CTO of Imagimob. “Compared to our previous workflow which was more top-down with a fixed pipeline, visualizing the entire workflow as a graph allows for a better overview of your process while unlocking the possibility to combine and test things in new and different ways—and ultimately build better models and accelerate time to market."
He added, "Moving over to the Graph UX and its capable ML backend is a big change for us and our users. We are taking a holistic approach to ML modeling which, in our industry, is unique.”
Imagimob Studio’s recently introduced Graph UX provides a visual overview of the modeling canvas, with the ability to zoom in and out and work at different levels of complexity. In addition to making it easier to work more efficiently as a team, Graph UX makes ML modeling projects more accessible to those with varying skill sets and experience levels. Furthermore, it can work at different levels of complexity and abstraction in the same workspace.
Imagimob Studio’s Graph UX update features include:
- Built-in data collection.
- Real-time model evaluation.
- Ability to evaluate and run multiple models in parallel or in sequence.
- Post-processing of model outputs.
- Support for Infineon boards right out of the box, and support for more hardware will be added soon. Users can also integrate their own hardware.
- Supports advanced models and hardware with multiple accelerators and processing units, such as Infineon’s PSoC Edge.