Introducing Palette Edgematic’s no code edge ML Development

SiMa’s no code, drag and drop programming environment for edge ML application development, enables the masses to visually create, build and deploy edge ML solutions on SiMa’s MLSoC silicon.

Evaluate Palette Edgematic free by joining our Developer Site

Palette Edgematic transforms the ML developer’s user experience in creating and analyzing edge ML pipelines using a drag and drop approach. Palette software then builds an execution pipeline with the click of a button that programs multiple processors executing on the MLSoC silicon. The Edgematic Graphical User interface (GUI) provides a no code methodology for creating edge ML designs. The Palette core software can then build and deploy this resulting code to a remotely accessible edge ML client device.


  • Compile and evaluate Any ML model, from framework to silicon without coding!
  • Build and evaluate Any pipeline, using python code to silicon with limited coding!
  • Deploy and manage edge ML applications executing on silicon
  • Support python and full Gstreamer pipeline development using plug-in libraries
  • Customize embedded Linux run-time environment for hosting edge ML applications
Palette Edgematic Functional Description
Palette Edgematic Canvas changes the entire approach to effortless ML edge development by delivering a no code, visual programming tool with Palette Edgematic. Using a canvas to graphically drag and drop models and processing elements provides a developer with a methodology to create a computer vision pipeline in minutes. Using this Evaluator Edition, Edgematic users evaluate the extensive SiMa library of ML models and application pipelines with push button compilation and deployment capability. The pipelines are built using the Model Library and pre and post processing plug-ins from the catalog in the associated window in the Edgematic Canvas. This no code capability extends to plug-ins for data source and sinks to run data sets through actual edge devices, evaluating their KPIs, and then modifying and iterating these pipelines without detailed embedded programming.

Model Library

Palette Edgematic provides a reference library of models targeting common ML network models that are quantized, compiled and optimized by the Edgematic compiler. Execute over 40 different models covering object detection, tracking, classification, then evaluate their Key Performance Indicators (KPIs) in minutes on actual silicon.

Example ML Networks Supported
  • ResNet
  • VGG
  • Inception
  • Squeezenet
  • Resnext
  • MobileNet
  • Centernet
  • reid
  • densenet
  • rcnn
  • human pose estimation
ML Application Graphical Development
  • End-to-end pipeline performance analysis
  • Analyze your target application running on the MLSoC
  • Library of Computer Vision (CV) plugins
Pipeline Visualization Interface
Application Library

Palette provides a reference library of application pipelines targeting common use cases. These include:

  • Face mask detection
  • People tracking
  • Open-pose
  • Open-cv-pipeline
Deploy and run an ML Model on a remote device
MLSoC Device Management Interface

A key component is the deployment and device management interface. Palette Edgematic provides a deployment dropdown menu of available remote devices that Edgematic can download the ML application pipeline executables to the board, manage their execution and obtain results from the ML pipeline execution. In this version, SiMa has provisioned a set of MLSoC boards accessible from the cloud that customers can be provisioned to access as part of an evaluation on silicon.

Measuring Key Performance Metrics:

The secure connection that manages the remote edge MLSoC device is utilized to monitor the execution, extract metrics for the KPI analysis. The remote device receives a data stream, processes the pipeline and returns results. During this process, the execution pipeline resources are monitored and presented graphically to identify the pipeline utilization of the Computer Vision Unit(s) (CVU) contained four processor cores, the Machine Learning Accelerator (MLA) and the Application Processor Unit (APU) containing a quad ARM processor subsystem. The system will identify the frames per second throughput of each subsystem to identify overall performance pipeline congestion.

  • Visually analyze multi-core utilization
  • Utilize for efficient pipeline mapping
Key Performance Indicator (KPI) Interface
How does Palette Edgematic help developers today?

Palette Edgematic platform provides:

  • No code computer vision pipeline creation, evaluation and iteration creating faster time to value. Visually Create, Build, and Deploy in minutes without embedded programming skills. This is a key value proposition since it expands the ability of a wide range of developers, data scientists, and machine learning developers without embedded programming experience to actually run their pipeline on the edge in real time with actual target data to assess the performance, accuracy and model performance in a real world scenario.
  • Versatility. Tackle any model, any computer vision problem imaginable. Auto-partition and compile across MLA and ARM processors.
  • Simplicity. Simplifying the development of computer vision pipelines with a graphical user interface to make this process accessible to all.
  • Automation. Eliminating the need for hand coding with effortless push button compile, build and deploy.
  • Performance. Exponential performance gains beat legacy solutions designed for the data center.

Evaluate Palette Edgematic free by joining our Developer Site

To receive more information on Palette Edgematic Software, please fill out the form and our team will get back to you.