Things to keep in mind when labeling data for Machine Learning/Deep Learning | Coursera Community
Coursera Header

Things to keep in mind when labeling data for Machine Learning/Deep Learning

  • 10 August 2019
  • 0 replies
  • 70 views

Badge +1


Hi everyone.
I am doing the deep learning specialization and choose to do an Final Year Project (FYP) of my computer science degree related to it. Deep learning would not be possible without all the data and compute power we have today.

For my FYP, I've had to label my own data since no public dataset was available for road cracks that met our needs. So much of my study was focused on the models, the architecture and the use of those algorithms. I had no clue on how to label my own data.
I think the community would benefit if any experienced annotator writes a piece on this topic. I thought of writing one but I am far from experienced at this point. I keep learning new things everyday.
I tried LabelMe, LabelBox and finally Supervisely. The screenshot is an example of my work. I've used the brush tool to annotate the cracks with class 'Crack' which I choose to create.

0 replies

Be the first to reply!

Reply