This is a question about how to label motorcycles for transfer learning

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

-Hardware Platform : 8GPU SERVER SYSTEM G482-Z54
-TAO-5.0
-GPU: A100

To conduct transfer learning on motorcycles, we collected 5,000 images from the field during the day, night, and rainy days.
When labeling a motorcycle with a person riding in an image, should I draw the bbox including the person or should I label only the motorcycle without the person?

3

Moving to TAO forum.

4

It depends on your target. If the task is going to detect a motorcycle only, then label only the motorcycle without the person. If the task is going to detect a person only, then label the person only.

5

All right. thank you

2023년 10월 12일 (목) 오후 6:39, Morganh via NVIDIA Developer Forums <notifications@nvidia.discoursemail.com>님이 작성:

6

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.