MuseBox AI on steroids! On-premise Demo with Xilinx AMD FPGA

MuseBox is the MakarenaLabs flagship product and we make it available to you a new on-premise Demo on the MuseBox site. We show seamless integration of AI applications simultaneously on several different Xilinx AMD hardware. From Ultra96 and Kria, U200 and U50 Alveo cards up to VCK5000. You can execute AI inference from your web browser wherever you are, on any of the AI models we made available, running them in real-time on the boards and getting performance results.

What is MuseBox?

MuseBox ( ) is a production system for extracting data and information from audio and video sources using artificial intelligence. It works on every Xilinx AMD FPGAs. You can compose the entire smart system according to your desires, or visit the store ( ).

The neural networks included in the MuseBox system are created from scratch, by the MakarenLabs experts' team, leveraging the AMD DPU, and are completely customizable.

MakarenaLabs team working on real-time inference acquired over years of deep knowledge on using FPGA technology in artificial intelligence and machine learning, offering custom AI models. 

Reason for the MuseBox Demo

Today chip shortage is affecting every company, but at the same time, you need to have an easy and eloquent way to understand the capability of the FPGA hardware to run AI. Thus we thought to make it available by virtualizing MuseBox. Real hardware you can run everywhere and every time.

Supported boards

The currently supported boards are: 

  • KRIA SOM KV260, ZCU104, CorazonAI from iWave, and Ultra96v2 from AVNET for embedded cloud systems;
  • Alveo U50, Alveo U200, VCK5000 for on-premise, mini data centers, micro-data centers.

Look at the video:


Test it with your images!

  1. Access the demo from our webpage (free registration required) and log in
  2. Then, enable your webcam, (it might be already enabled on your PC)
  3. After that, take a photo or upload a photo on the site
  4. Following, select the board where you want to run the demo
  5. Next, select one of the artificial intelligence tasks
  6. The inference will start, it will take a while depending on your internet connection, but the measures we show are those from the instant we get the picture on the real board, and the site page will show you the result drawn on the original picture, plus the performances in terms of execution time, frames per second, and energy used.
  7. A JSON response from the system, with the API structure, will be produced as well.

Why is it useful?

  • You can test the FPGAs capability remotely, thus a fast evaluation process.
  • Then, It demonstrates that remote edge computing, proposed by MakarenaLabs, is a viable solution for accelerating processes and integrating the already-made and custom-made solutions with MuseBox
  • Finally, It uses your camera thus no need to install one, zero burden