Hi,
I am using jetson-inference/docs/pytorch-ssd.md at master · dusty-nv/jetson-inference · GitHub link and trained the custom object detection model with 4 classes each having different number of images e.g class-1 having 300, class-2 with 60, class-3 30 images and class-4 having 15 images.
While testing the model, model is detecting almost all object as class-1 even if the object belongs to class-2,3 or 4.
I have trained model with argument --balance-data also. Still getting same results.
How can I train the model using other ways.
Please help me out!
Hi,
Are you able to increase the images of each class to at least 100 and try it again?
Thanks.
no. right now that is not possible. I have to do the training with available images only. Is there any other method to balance the data.
Hi,
Please search for some data augmentation research to see if they can solve your issue.
Thansk.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.