Submission Requirements

We have released the evaluation project for teams to evaluate the three metrics on their end. The evaluation project can be accessed here ( In the evaluation project, the four metrics (i.e., F_beta, latency, Generalization score, and flash usage) will be evaluated based on the performances obtained from the targeting MCU.

You can submit your design using the submission links below. Submit all source files for your design in a zip archive. In the zip file, you should submit the following:

  • The repo contains the model training design (normally in python) using frameworks such as PyTorch and TensorFlow.
  • In your C source code repo, you should modify the corresponding C source files based on the instructions provided in evaluation code to enable the evaluation. For example, if you are using X-CUBE-AI, in the C source code project generated by CubeMX, you should replace 4 files indicated here and include the whole project in your submission.
  • The trained model weights file in any format such as ONNX, tflite.
  • to briefly illustrate the requirements and how to run your code.
  • Submission Links

    Final Submission (by October 8th)