Training and classifying on the background class

89 views
Skip to first unread message

Daven Savoie

unread,
Apr 30, 2020, 10:09:07 AM4/30/20
to agriculture-vision
Hello everyone, 

I am a data science student working with this dataset. I have been struggling to understand how to create the label for background. I have been able to generate predictions for all other classes with 3 different but am unsure how to identify the background as a unique class. I have been searching google but have not found solutions, I would love to see my test score but without a background category, i cant train to predict it.

Could someone explain how to create and work with the background class? How do i train the data to identify a category that is not explicitly labelled?

Let me know if you need clarification on my question, I am an intermediate level learner and this is my first real world recognition problem.

Many thanks,
Daven

Bingchen Zhao

unread,
May 1, 2020, 3:20:50 AM5/1/20
to agriculture-vision
Hi Daven,

Our team ranked 8th in the challenge, our code is published on github, https://github.com/LAOS-Y/AgriVision, I hope this can be useful for you

Best,
Bingchen Zhao

在 2020年4月30日星期四 UTC+8下午10:09:07,Daven Savoie写道:
Reply all
Reply to author
Forward
0 new messages