Coding Period: Week 3

less than 1 minute read

Overview

So basically, the work in the third week was just an extension to the work in the Second week, where the main idea was to explore different Models available in the ONNX Model Zoo for the Exercise and to select the best one from the bunch.

Objectives

  • Explore the TFOD (TensorFlow Object Detection) Api and also netron.app
  • [] compare ONNX and NNEF using AlexNet
  • Try to find a better model for the RoboticsAcademy Exercise
  • Update blog and push everything to github

The execution

Again, since I had installed the TensorFlow Object Detection API in the second week, this week was relatively less heavy for me. After like reading about the Pros and Cons about the different models [5], I decided that we should use the SSD_Volumenet model for this purpose. Now, the thing is that we were previously using the SSD-VolumenetV1 for the Exercise already, and I proposed to maybe try the different versions of the same Model, which would be performed for the next week evualations.

References

[1] https://github.com/JdeRobot/RoboticsAcademy
[2] https://onnxruntime.ai/
[3] https://github.com/onnx/models
[4] https://www.tensorflow.org/tutorials\ [5] https://www.geeksforgeeks.org/difference-between-yolo-and-ssd/\

You May Also Enjoy