When it comes to training AI models for the Edge, bad data produces bad models
Failed to add items
Add to basket failed.
Add to Wish List failed.
Remove from Wish List failed.
Follow podcast failed
Unfollow podcast failed
When it comes to training AI models for the Edge, bad data produces bad models
-
Narrated by:
-
By:
About this listen
The Edge AI Foundation recently put together a challenge, proposing to Computer vision engineers, Data scientists, AI engineers and researchers to produce the best Vision AI models for the Edge using a large open source dataset. Participants were to approach the problem from the perspective of the models or from the perspective of the data.Kais Bedioui, Computer Vision engineer, was in the later camp and tells us how he used off-the-shelf AI models to clean up the large dataset, identifying non-labelled images or gaps and errors images annotations before even starting to train a model for the Edge.🔗 Check out Kai's project at https://github.com/kais-bedioui/Wake_Vision_Challenge_Data_Centric_Track🙏 This special episode is sponsored by the Edge AI Foundation https://www.edgeaifoundation.org/